# Falls River Media > Strategic content and practical AI automation for small teams and solo experts. Clear positioning, real writing, results you can measure. Public Ghost content for AI and LLM tooling. This file includes a bounded export of public pages first, then recent public posts. Append `.md` to any post or page URL to get the content in Markdown (for example, `/example-post.md`). ## Pages ### About URL: https://fallsrivermedia.com/about/ Last updated: 2026-07-07T02:41:18.000Z ### Small team. Strong current. Falls River Media exists for the businesses that big agencies overlook: solo experts, small teams, and founders who have something worth saying but not a marketing department to say it with. We do the kind of brand and strategy work you would expect from a much larger studio: the naming, positioning, and visual identity that set a business apart, and we deliver it fast. Then we build and run everything that carries it. Content, campaigns, events, and AI-powered systems work together, so a team of one or two can publish, promote, and follow up like a team of ten. One webinar becomes a month of content. One form submission becomes a researched, personalized report. One good idea travels every channel it deserves. ### What we believe **You are paying for honest judgment.** We are not yes-people, and if something you want isn't in your best interest, we will gently steer you toward something better. **People come before algorithms.** Our advice rests on the durable fundamentals of human relationships. Trends and platforms come and go; the person actually reading your message does not. **We are easy to work with on purpose.** We move fast, keep things simple, and never lock you in. The best work leaves you with more than you knew to ask for. ### Who you'll work with ![Lara Hill](https://storage.ghost.io/c/27/85/2785dbe7-4d85-45c4-81d8-7ac0f5cffe06/content/images/2026/06/lara-hill-headshot.png) #### Lara Hill Founder & Lead Strategist Fifteen years in marketing strategy for ecommerce and tech startups, with a creative eye for tasteful design. Lara shapes the branding, color, and voice of a business so it resonates with both the owner and the audience they want to reach, and she is a longtime advocate for AI literacy and ethical AI. The original photography across our site is her own. ![Noel Projimo](https://storage.ghost.io/c/27/85/2785dbe7-4d85-45c4-81d8-7ac0f5cffe06/content/images/2026/06/noel-projimo.jpeg) #### Noel Projimo Media Production Specialist Noel turns ideas into things people can see and share: video and social content built to carry a message across every channel. He brings more than a decade of remote production and marketing support, and has collaborated with Lara across many projects over the years. ![Khalil Gueye](https://storage.ghost.io/c/27/85/2785dbe7-4d85-45c4-81d8-7ac0f5cffe06/content/images/2026/06/khalil-gueye.jpeg) #### Khalil Gueye Automation & Systems Developer Khalil builds what runs behind the scenes: the automations, tools, and data workflows that let a small team move like a large one. His background in economics and cybersecurity gives him an eye for both the numbers and the safeguards, so the technology stays useful and trustworthy. ![Maychil Projimo](https://storage.ghost.io/c/27/85/2785dbe7-4d85-45c4-81d8-7ac0f5cffe06/content/images/2026/06/maychil-projimo.jpeg) #### Maychil Projimo Operations & Content Specialist Maychil keeps the work moving: content production, research, and the day-to-day operations that hold a project together. With experience spanning web, design, and language instruction, she brings precision and a service-first approach to everything she touches. ### How we work Most clients start with one of three doors: 1. **A project:** a webinar series, a launch, a content backlog that needs clearing. 2. **A system:** an AI-powered workflow we design and hand over (or run for you): lead magnets, content repurposing pipelines, automated research and reporting. 3. **A partnership:** ongoing marketing support, from strategy through execution, sized for a small-business budget. Not sure which? Tell us what you're working on and we'll help you find the right fit. And if you want to see what we build first, try our [Speaker Opportunity Finder](https://fallsrivermedia.com/speaker-opportunity-finder/): tell it your expertise and it emails you a personalized pipeline of speaking opportunities. That's the kind of leverage we put inside your business. ### Let's talk Tell us what you're trying to say, and to whom. We'll tell you honestly whether we can help, usually within one business day. [Tell us about your project](https://fallsrivermedia.com/contact/) ### Services URL: https://fallsrivermedia.com/services/ Last updated: 2026-07-07T02:41:17.000Z Falls River Media gives small teams and solo experts agency-caliber brand and marketing work, delivered fast. We start with strategy, your brand, your positioning, your message, and then build and run everything that carries it: content, campaigns, events, and AI-powered systems that do the work of a full department. ### Brand & Positioning **Know who you are, then say it everywhere.** Naming and brand architecture, positioning and messaging, and a complete visual identity: logo, color, and type. We define the story first, then the look and voice that fit both you and the people you want to reach, and we carry it across every page, deck, and channel. ### Virtual Event Production **Professional support for webinars and livestreams.** End-to-end event management, speaker preparation and coaching, technical hosting and production, and post-event reporting for smooth, stress-free virtual delivery. ### Content Development **Content that extends your reach.** Webinar and event repurposing, blog articles and thought leadership, LinkedIn and social posts, lead magnets, and email nurture campaigns. ### AI Automation Systems **A marketing department's output, without the headcount.** We design AI-powered workflows you own: lead-magnet engines, content repurposing pipelines, automated research and personalized reporting, and follow-up systems that never forget. Built on practical tools, documented, and handed over, or run for you. ### Campaign Support **Consistent promotion across channels.** Email marketing setup and execution, social scheduling, content calendars and campaign coordination, list segmentation and audience targeting. ### LinkedIn Strategy & Support Optimize your LinkedIn presence with profile improvements, livestream guidance, and audience growth strategies. We help you position your expertise, increase visibility, and build a stronger following. ### CRM & Marketing Systems As a certified HubSpot Solutions Provider, we help you choose and run the right-sized stack: CRM setup, list management, email automation, landing pages, and reporting. And when a $20/month platform is more than you need, we'll tell you that, too. --- ### Let's talk about your next project Whether you're building your first webinar series or scaling your content marketing, we can help you move quickly and stay focused on results. ### Speaker Opportunity Finder URL: https://fallsrivermedia.com/speaker-opportunity-finder/ Last updated: 2026-07-07T02:41:19.000Z Free Personalized Speaking Pipeline # Find the stages that are *looking for you* Tell us your expertise and the audience you want to reach. We'll match you against upcoming conferences and open calls-for-speakers, then email you a personalized pipeline of events worth pitching, with deadlines and submission links. [Get my Speaking Pipeline](#form) No cost. Takes 3 minutes. Pipeline delivered to your inbox. 1 ### Tell us about you Your topics, credentials, and the audiences and industries you want in the room. 2 ### We run the match Your profile is scored against our continuously updated database of conferences and open CFPs. 3 ### Get your pipeline Your ranked pipeline lands in your inbox: best-fit events, submission deadlines, links, and what to do first. ## Build my speaker pipeline Fields marked \* are required. First name \* Last name \* Email \* Speaking topics / areas of expertise \* Target audience \* Industries of interest Corporate / HR Education Healthcare Faith-based Government Tech Nonprofit Other What kind of speaking opportunities are you looking for? (select all that apply) Editorial: selected via submission (typically unpaid) Paid speaking (honorarium/fee) Open to sponsored slots (you pay to speak) Format preferences Keynote Breakout / workshop Panel Virtual In-person Speaking experience Just getting startedA few talks a yearRegular speakerProfessional / keynote speaker Travel region Anywhere in the U.S.Virtual onlyInternational LinkedIn profile (optional) Would you like follow-up from Falls River Media? Yes, help with speaker submissions Yes, broader marketing support No thanks Is there anything else we should know? (optional) Also send me *Undercurrents*, thoughts on message, media, and momentum. Get my Speaking Pipeline → We'll email your Speaking Pipeline and occasional speaking-industry insights. Unsubscribe anytime. ✓ ## Your Speaking Pipeline is being built We're matching your profile against upcoming events and open calls-for-speakers right now. **Check your inbox**. Delivered within 1 business day. Inclusion of an event in your pipeline is not an endorsement. Events are not vetted by Falls River Media. Always evaluate fit before pitching. ⏱ Delivered within 1 business day ## What you'll receive A ranked, deadline-ordered pipeline, the same format we build for our private speaker-placement clients. Speaking Pipeline Sample Snapshot Pipeline Top Shortlist Watchlist | Tier | Event | Audience | Deadline | | ---- | --------------------------------------------------- | ---------------------------------- | ----------------- | | 1 | **CFX Safety Conference**Birmingham, AL · Sept 2026 | 2,000+ facility & security leaders | Rolling. Act now | | 1 | **PMI Global Summit**Detroit, MI · Oct 2026 | PMO directors, delivery execs | Confirm this week | | 2 | **SHRM Annual Conference**Las Vegas, NV · June 2027 | HR & people leaders | Est. Oct 2026 | | 2 | **The AI Summit New York**NYC · Dec 2026 | Enterprise tech buyers | Jul 15, 2026 | \+ 26 more events, fit rationale, submission links, and a next-cycle watchlist in your full pipeline. **20–40**matched events per pipeline **12 mo**deadline look-ahead **Free**no card, no catch ### Legal URL: https://fallsrivermedia.com/legal/ Last updated: 2026-07-07T02:41:20.000Z Please read these Terms of Use ("Terms") carefully before using the Falls River Media website (the "Site"). ### Acceptance of Terms By accessing or using the Site, you agree to be bound by these Terms. If you do not agree to these Terms, please do not use the Site. ### Use of Site Content All content on this Site, including text, images, graphics, and logos, is the property of Falls River Media unless otherwise noted. You may not reproduce, distribute, modify, or create derivative works from any content without prior written consent. ### Disclaimer The information on this Site is provided for general informational purposes only and is not professional advice. We make no guarantees as to the accuracy, completeness, or suitability of the information. ### Limitation of Liability Falls River Media is not liable for any direct, indirect, incidental, or consequential damages resulting from your use of the Site. ### Third-Party Links The Site may contain links to third-party websites. Falls River Media is not responsible for the content or privacy practices of any third-party sites. ### Governing Law These Terms are governed by the laws of the State of North Carolina, without regard to its conflict of law provisions. ### Changes to Terms We reserve the right to modify these Terms at any time. Changes will be posted on this page, and continued use of the Site constitutes your acceptance of the revised Terms. ### Questions? Get in touch and we are happy to help. ### Privacy Policy URL: https://fallsrivermedia.com/privacy/ Last updated: 2026-07-07T02:41:19.000Z Falls River Media ("we," "us," or "our") respects your privacy and is committed to protecting any personal information you provide through our website. ### Information We Collect We may collect personal information such as your name, email address, company name, and phone number when you: - Submit a contact form - Subscribe to our newsletter - Request information or services ### How We Use Information We use the information you provide to: - Respond to inquiries and provide requested services - Improve our website, services, and customer experience - Send occasional marketing or informational emails (you may opt out at any time) ### Information Sharing We do not sell, rent, or share your personal information with third parties except: - As required by law - To trusted service providers who help us operate our business (and who are bound by confidentiality agreements) ### Cookies and Analytics Our website may use cookies or similar technologies to improve user experience and analyze site traffic. You can adjust your browser settings to disable cookies if you prefer. ### Data Security We take reasonable measures to protect your personal information from unauthorized access, alteration, or disclosure. ### Your Rights You may contact us at any time to: - Request access to your personal information - Request correction or deletion of your personal information ### Contact Information If you have any questions, concerns, or requests related to your personal data, please reach out and we will be happy to assist. ### Policy Updates We may update this Privacy Policy periodically. Effective date of the current policy: June 12, 2026. ### Contact URL: https://fallsrivermedia.com/contact/ Last updated: 2026-07-07T02:41:18.000Z # Contact Have a question, a project in mind, or just want to say hello? Send a note below and it comes straight to Lara. We typically reply within one business day. First name Last name Email Message Also send me *Undercurrents*, thoughts on message, media, and momentum. Send message Your message is sent to Falls River Media. We'll only use it to reply to you. ## Thanks — your message is on its way. We've received your note and will get back to you within one business day. ### Home URL: https://fallsrivermedia.com/home/ Last updated: 2026-07-22T12:11:01.000Z # Small team. *Strong current.* Brand, content, and AI-powered systems for solo experts, small teams, and founders who want strategy-led marketing without an agency in the way. [See what we do](https://fallsrivermedia.com/services/) [Read Undercurrents →](https://fallsrivermedia.com/undercurrents/) What we make ### Brand & Positioning Naming, brand architecture, positioning, and visual identity that signals what only you can deliver. [Brand work →](https://fallsrivermedia.com/services/) ### Content & Campaigns Editorial strategy, content production, and campaign systems that build audiences and reach buyers. [Content work →](https://fallsrivermedia.com/services/) ### AI Automation Systems Custom AI-powered automations that turn busywork into background work so small teams can ship like big ones. [Automation work →](https://fallsrivermedia.com/services/) Free 2-minute assessment ### Is your marketing keeping up with your business? Answer eight quick questions about your email, social, content, and tools. Get a tailored list of cost-efficient tools to consider, built on an SEO and AEO foundation. [Take the assessment](https://fallsrivermedia.com/marketing-stack-assessment/) Latest from Undercurrents [See all →](https://fallsrivermedia.com/undercurrents/) Loading… ## Have a project in mind? Tell us what you're trying to say, and to whom. We'll tell you honestly whether we can help, usually within one business day. [Tell us about your project](https://fallsrivermedia.com/contact/) ## Read *Undercurrents* Thoughts on message, media, and momentum, delivered by email. [Subscribe](#/portal/signup) ### Marketing Stack Assessment URL: https://fallsrivermedia.com/marketing-stack-assessment/ Last updated: 2026-07-21T20:15:31.000Z Free Assessment # Is your marketing keeping up with your business? Eight quick questions about your email, social, content, and tools. Takes about two minutes. You'll get a snapshot of where you stand plus a **tailored list of cost-efficient tools** to consider, built on an SEO and AEO foundation so your content works in both traditional search and AI answers. Start the assessment ## ← Previous question Your Results **Start with the foundation: SEO and AEO.** Before adding tools, make sure your content can be found. That means search engine basics (Google Search Console, a clear site structure) plus Answer Engine Optimization, so AI assistants like ChatGPT, Claude, and Perplexity can cite you when buyers ask questions. Every recommendation below builds on that base. ### Want your results and tool plan in your inbox? We'll send this as a one-page PDF you can keep and share. Email me my results By submitting, you agree to receive your results and occasional marketing tips from Falls River Media. Unsubscribe anytime. ### Want a hand implementing this? Falls River Media builds AI-augmented marketing systems for B2B and professional services firms. We don't just recommend tools, we set them up and train your team. [Book a free consult](https://www.fallsrivermedia.com/contact?ref=fallsrivermedia.com) Retake the assessment ## Posts ### Can I Use Lovable for Blogging? URL: https://fallsrivermedia.com/can-i-use-lovable-for-blogging/ Last updated: 2026-07-22T14:09:58.000Z Last month I wrote about [whether Lovable sites are findable by ChatGPT and Claude](https://fallsrivermedia.com/is-your-lovable-site-findable-by-chatgpt-and-claude/). That post answered a technical question, and it ended on an optimistic note: Lovable fixed its biggest visibility problem in May, and the platform keeps getting better. But there’s a question sitting right behind that one. If you love building on Lovable (and having built on it, I understand why), can it carry your blog too? Before answering, it’s worth stepping back to a more basic question: why blog at all? For a business, a blog usually exists to do three jobs. 1. **Bring in organic traffic.** Someone searches, or asks an AI assistant, and your article is the answer. They had never heard of you until the moment you were useful to them. This is the job most content strategies are actually betting on. 2. **Reach your subscribers.** A new article goes out to the people who already said yes to hearing from you, and keeps the relationship warm between (or before) purchases. 3. **Feed your social presence.** Each article gives you something substantive to post, and when the link gets shared, it unfurls into a proper preview instead of a bare URL. If a blog is doing those three jobs, it’s working. If it isn’t, what you have is writing on your website. So the real question isn’t whether Lovable can build you a blog. It can, in an afternoon, and it will look great. The question is whether that blog does the three jobs. Let’s take them one at a time, starting with the one that has the biggest catch. ## Job 1: getting found, and the sitemap problem Lovable is an app builder. When you prompt “add a blog to my site,” it builds you one from scratch, to your description, and every feature a blog needs exists only if you asked for it. Most of what makes a blog findable is invisible, which is why this is where that arrangement bites hardest. Here’s what has to happen, every single time you publish, for a post to get found. The post has to land in your sitemap, the list of your URLs that search engines read. The page needs its own title, description, and canonical tag, done correctly. Search engines want structured data identifying it as an article. If AI visibility matters to you, there’s also llms.txt, a newer convention that hands AI assistants a clean map of your site’s content. On a publishing platform, all of that happens by itself the moment you hit publish, every time, and you never see any of it. On Lovable, each item happens only if you do it, or only if you’ve had a system built that does it for you. So the question that decides whether blogging on Lovable is sustainable isn’t “can it do SEO.” It’s “am I going to run this checklist by hand every single time I post?” The sitemap is the sharpest example, so let me stay on it a moment. Google’s own guidance says a sitemap matters most for exactly the kind of site a solo owner or small team runs: newer, with few external links pointing at it, where crawlers may never stumble onto your pages on their own. Publishing platforms have handled this forever: WordPress has generated its sitemap automatically since 2020, covering every published post and page by default, and HubSpot’s documentation says it plainly: HubSpot “automatically adds your live HubSpot-hosted website pages and blog posts to your sitemap.” You publish, the sitemap knows, and you never think about it. On Lovable, unless you’ve built something better, the sitemap is a file the AI wrote once, listing the pages that existed on the day you asked for it. Publish a post next month and nothing adds it. This isn’t hypothetical. I recently reviewed a Lovable-built site with a truly strong article on it, published in the spring, with the visible work all done: clean copy, careful per-post metadata. But the site’s public sitemap was months old and the article wasn’t in it, and Search Console filled in the consequence: eleven weeks after publication, Google had never crawled the article once. Not ranked poorly. Never seen. Nothing on the site looked broken, and nobody had been careless. The sitemap just isn’t a file anyone is told to watch. (It can drift the other way too: because prompt-written sitemaps are AI-written text rather than something generated from the site’s structure, they’ve been documented both listing pages that don’t exist and leaving out pages that do.) Notice what kind of failure this is. A broken layout announces itself the moment you look at your site. A stale sitemap announces itself never. Everything looks fine, every post looks live, and months of publishing quietly builds less audience than it should. Care doesn’t protect you here, because you can do every publishing task you can see and still miss the pipes that only a publishing platform would have remembered on your behalf. Now the good news, because there is real good news: the answer to “will I do this every time?” can be no. You just have to ask Lovable for the right thing. There’s a big difference between “add my new post to the sitemap,” a chore you’d be repeating every week, and “rebuild my sitemap automatically from my site’s actual pages on every build,” a system you set up once. The same move works down the rest of the checklist. Article pages can be built to pull their titles, descriptions, and structured data from the article itself, instead of being hand-decorated one at a time. An llms.txt file can be generated from that same source, so it never falls out of sync. I’ve seen a Lovable site wired exactly this way, and it holds up. The principle: make the site remember, so no person has to. Two caveats before you relax. Setting the system up takes a focused session and some credits, and you have to know to ask, because nothing in the platform will suggest it. And once it’s built, trust it the way you’d trust any plumbing: verify from the outside. After your next post goes up, load your live sitemap and look for the new URL, then confirm in Google Search Console that the post gets indexed. Search Console is free, and it’s the closest thing to ground truth on whether any of this is working. ## Job 2: reaching the people who already said yes The second job of a B2B blog is feeding your email list, and here the gap is easier to see once you know to look. Lovable has no built-in newsletter. No subscriber list, no signup form wired to anything, no send button. If your content rhythm is “new article goes to subscribers,” that whole chain is yours to assemble: an email service, a signup form connected to it, and a working connection that tells the email service a new post exists. Some publishing platforms bundle that entire chain, so emailing subscribers is simply part of hitting publish. On Lovable, each link exists only if you built it, and the failure mode is the same silent kind as the sitemap: nothing on your site looks broken when a link is missing. Your new post simply goes out to no one until you share it by hand. ## Job 3: giving you something worth sharing The third job is social. The article itself is the substance, and Lovable has no bearing on whether your thinking is good. What it does bear on is the moment your link lands on LinkedIn: whether it unfurls into a title, description, and image, or sits there as a bare URL. Those previews come from per-post metadata, which Lovable will happily add when asked and will not maintain on its own. Edit a headline later and the preview metadata keeps the old one unless someone updates it too; I’ve seen exactly that drift on a live Lovable site. The check is quick, at least: paste a post link into LinkedIn’s Post Inspector before you share, and you’ll see what the platforms see. ## The everyday frictions Two more things belong in the decision, briefly. Publishing costs credits. Lovable charges per AI message, roughly half a credit for a small tweak and about two for a full page, so the post itself, the revision, and the typo fix each have a meter running next to them. The dollars are small at a modest cadence. The subtler cost is that a meter on your most frequent action teaches you to hesitate before fixing small things, and that’s a bad habit to train into a writer. And your words live inside your app. Depending on how the AI set things up on day one, posts end up written into the site’s code, stored as files in the project, or kept as rows in a database. Files and database rows can be exported and moved. Posts woven into the code have to be pulled loose one at a time. Two years of writing is an asset, and how portable it is gets decided at the start, usually without anyone noticing a decision was made. ## So, can you? Yes, you can blog on Lovable. Whether you should depends on which of those three jobs your business is counting on. If your site is really an app (a tool, a calculator, a booking flow, a community product) and the blog is a supporting act, go ahead. Ask for the sitemap that rebuilds itself, the subscriber plumbing, and the per-post metadata up front, run the checks above once, and enjoy having everything in one place. If organic traffic, a subscriber list, and a steady social presence are the point, look hard before you commit, because those are exactly the three places where a Lovable blog needs you to notice what’s missing. None of it is impossible to add. All of it fails silently when you don’t. And if you’re already publishing on Lovable, don’t take my word for any of this. Open your own sitemap and look for your newest post. If you have a subscriber signup, test it end to end. Paste your last article into a link inspector. Twenty minutes, and you’ll know which of the three jobs your blog is actually doing. I’d love to hear what you find, especially if you’re mid-decision right now. *Facts checked as of July 2026\. Lovable ships changes quickly, so some details may shift after this date.* ### Sources - [Google Search Central, sitemaps overview](https://developers.google.com/search/docs/crawling-indexing/sitemaps/overview?ref=fallsrivermedia.com) - [Make WordPress Core, XML sitemaps in WordPress 5.5](https://make.wordpress.org/core/2020/07/22/new-xml-sitemaps-functionality-in-wordpress-5-5/?ref=fallsrivermedia.com) - [HubSpot Knowledge Base, view and edit a HubSpot-hosted domain sitemap](https://knowledge.hubspot.com/domains-and-urls/view-and-edit-a-hubspot-hosted-domain-sitemap?ref=fallsrivermedia.com) - [Encited, How to generate a sitemap with Lovable reliably](https://encited.com/blog/how-to-generate-sitemap-on-lovable?ref=fallsrivermedia.com) - [Lovable pricing, credit examples and plans](https://lovable.dev/pricing?ref=fallsrivermedia.com) - [Falls River Media, Is Your Lovable Site Findable by ChatGPT and Claude?](https://fallsrivermedia.com/is-your-lovable-site-findable-by-chatgpt-and-claude/) ### ChatGPT Ads Just Opened Up. Here's What HubSpot's Integration Actually Gives You. URL: https://fallsrivermedia.com/chatgpt-ads-hubspot-which-tier-you-need/ Last updated: 2026-07-15T12:24:31.000Z In February, running an ad inside ChatGPT required a minimum commitment somewhere north of $200,000\. As of May, it requires nothing. And as of last week, you can manage the whole thing from inside HubSpot. That is a fast collapse, and it has produced a lot of breathless coverage. Most of it skips the two questions that decide whether this channel is worth your time: who actually sees these ads, and what does your HubSpot subscription let you measure once the clicks start arriving. Here is the current state of both. ## What changed, in order ChatGPT's ad pilot launched February 9, 2026 for users on the Free and Go tiers, invitation-only, with a minimum spend reported between $200,000 and $250,000 and CPM-only bidding at a $60 default. That put it out of reach for everyone except holding companies and a handful of enterprise brands. It did not stay there long. By late March the pilot had reached Canada, Australia, and New Zealand, though accounts differ on whether those markets arrived at launch or a few weeks after it. In April the floor fell to $50,000 and CPC bidding entered limited pilot at a recommended $3 to $5 starting bid. Then on May 5, OpenAI opened the self-serve Ads Manager to every US business, removed the minimum spend entirely, and shipped a measurement pixel and Conversions API. Cost-per-action bidding followed across late May and early June for accounts with conversion tracking configured. Roughly a quarter of a million dollars down to zero in three months. CPMs compressed alongside the floor, from the $60 launch rate to somewhere around $25 to $60 depending on category. That is the normal pattern for a new ad market finding its clearing price. It is just happening faster than usual. ## What HubSpot added HubSpot's ChatGPT Ads integration went into beta in mid-July. It is not generally available yet, so customers have to request access. HubSpot consultant Carrie Gallagher published [the clearest walkthrough of the beta's mechanics](https://carriegallagher.com/hubspot-chatgpt-ads-integration/?ref=fallsrivermedia.com) so far, and the summary below follows her account alongside HubSpot's release notes. If you want the step-by-step, start there. What we want to add comes after it. Once you have access, you connect an OpenAI Ads account under Settings > Marketing > Ads using an API key from your OpenAI Ads dashboard. Campaign creation lives at Marketing > Ads, with a ChatGPT ad type that asks for a short title, body copy capped around 100 characters, an image, and a destination URL. Targeting is the genuinely different part. Instead of demographics or keywords, you supply context hints: free-form phrases describing the kinds of conversations where your ad belongs. You are not describing a person. You are describing a question. From there, ChatGPT campaigns appear in the same Manage and Analyze tabs as your other ad networks, auto-tracking appends UTM parameters, and contacts who click can enroll in workflows. One detail worth flagging: HubSpot's beta description specifies fixed CPM pricing, while OpenAI's direct platform now supports CPC and CPA bidding as well. If bidding model matters to your test design, confirm what the beta actually exposes before you plan around it. ## Which HubSpot tier you need This is where most of the coverage gets sloppy. The claim making the rounds is that the integration works on every Marketing Hub tier, including Free. That is true about access and misleading about value. Connecting an ad account is available on all HubSpot plans. HubSpot's own documentation confirms the ads tool is part of Marketing Hub and can be used for free. Measuring what happens afterward is a different question: TierWhat you get Free Last ad interaction attribution only Starter Last ad interaction attribution only, plus expanded audience options Professional All attribution report types, plus new deals and new customers metrics, plus the campaign Revenue report Enterprise Adds the campaign Revenue attribution report and multi-touch attribution models HubSpot states this directly in its documentation: accounts on the free CRM and Marketing Hub Starter have access to the Last ad interaction report only, while Professional and Enterprise accounts unlock the rest. The new deals and new customers metrics likewise require Professional or Enterprise. A note on the two rows at the bottom, because the naming is easy to misread. These are two different reports. The campaign **Revenue** report requires Marketing Hub Professional. The campaign **Revenue attribution** report, which is the one with selectable multi-touch models, requires Enterprise. Both appear in HubSpot's campaign documentation with those subscription requirements attached. Worth confirming which of the two you actually need before you budget for a tier. So the pitch you have probably read, that you can trace a click through to a contact, a deal, and revenue, describes the Professional feature set. On Free or Starter you get last-click and nothing downstream. You can absolutely run the ads. You just cannot answer the question you ran them to answer. **One caveat we cannot close for you.** HubSpot's ad network documentation still lists only Facebook, Google, LinkedIn, and TikTok. ChatGPT is not in the knowledge base yet, because the integration is days old and gated. The table above reflects the generally available ads tool. HubSpot has published nothing about whether the beta carries its own eligibility requirements, and betas are frequently gated separately. The request-access form is the only way to know. ## Who actually sees these ads Ads serve to ChatGPT's Free and Go tiers. Plus, Pro, Business, Enterprise, and Education subscribers do not see them, and OpenAI has positioned ad-free access as a permanent property of paid plans rather than a temporary state. That matters enormously depending on who you sell to. If your buyers are enterprise employees working inside company-provisioned ChatGPT seats, you are structurally unable to reach them here, no matter how good your context hints are. If your buyers are individual practitioners researching a decision on their own account, the picture is very different. This single fact does more to determine fit than any performance benchmark. Work it out before you look at CPCs. ## What the performance data does and does not say You will see one number repeated everywhere: a 0.91% click-through rate against a 6.4% Google Search benchmark, roughly seven times lower. Treat it carefully. That figure traces back to a single Adthena client, in one sector, during the managed-pilot phase in March, when a reporting glitch was also preventing some advertisers from seeing their own campaign data. Adthena's own CMO framed it as a cautionary signal, not a benchmark. Every article citing it is citing the same data point. Other early reads point elsewhere. One performance marketing agency reports CTRs ranging from 0.8% to 2.3% depending on vertical, with education among the stronger fits. Another published 15 days of live account data showing CPC around $1.72, well under the widely quoted $3 to $5\. Criteo reports LLM referrals converting at 1.5x other channels, though that comes from its own retail client base and has not been independently verified. Notice what all of those have in common. Each is one advertiser's account, shared voluntarily, in a category that may look nothing like yours. OpenAI has released no aggregate cross-advertiser data at all. So there is no benchmark here to hit or miss, only a scattering of other people's results that happen to be the only numbers anyone has. Read them as weather reports from somewhere else, not as a forecast for your own campaign. ## What we would tell a client Three things. Check the audience question first. If the people who buy from you are sitting on paid ChatGPT seats, stop here. The rest of the analysis does not matter. Check your tier second. If you are on Free or Starter, you can run the ads but you will be flying on last-click. Decide whether that is enough to learn something, or whether the test is only worth running from Professional up. Size the test for learning, not for revenue. The channel is too young and the data too thin for anyone to promise you a return. What an early test does buy is a real understanding of how your buyers describe their problems inside AI, and that intelligence is useful well beyond this one ad platform. The floor came down. That does not mean the answer is yes. It means the question is finally worth asking. --- *Falls River Media builds AI-augmented marketing systems for B2B and professional services organizations. If you are weighing whether a new AI channel belongs in your mix,* [*get in touch*](https://fallsrivermedia.com/contact/)*.* ### How to Track AI Referral Traffic in GA4 in 2026 URL: https://fallsrivermedia.com/how-to-track-ai-referral-traffic-in-ga4-in-2026/ Last updated: 2026-07-12T15:18:14.000Z Your website is already getting traffic from AI tools. The good news is that, as of May 2026, Google Analytics 4 will finally show you some of it without any setup. The catch is that the native view captures only part of the picture. When a potential client asks ChatGPT to recommend a firm in your category and your company comes up in the answer, that person often clicks through to your site. For years, that visit got buried inside GA4's generic "Referral" bucket, or worse, dumped into "Direct" as if they had typed your URL from memory. You had no clean way to see it. In May 2026, Google changed that. GA4 now has a built-in AI Assistant channel that automatically separates a chunk of this traffic out for you. This guide covers what that native channel does, the significant gaps it leaves, and how to build a complete view, including a way to have Claude set the whole thing up for you. ## What GA4's New AI Assistant Channel Does On May 13, 2026, Google added a native AI Assistant channel to GA4's Default Channel Group, with broad availability reaching most properties by early June. It requires no configuration. When GA4 recognizes a visit referred by a supported AI assistant, it automatically tags that session and slots it into the AI Assistant channel in your standard acquisition reports. To find it, go to Reports > Acquisition > Traffic Acquisition and set the primary dimension to Session default channel group. You should see "AI Assistant" listed alongside Organic Search, Direct, Social, and Referral. For a fast read on whether AI tools are sending you traffic, this is genuinely useful, and it took zero effort on your part. But treat the number as a floor, not a ceiling. ## What the Native Channel Misses (This Is the Important Part) The native channel cleans up the easy cases. It does not solve the underlying measurement problem. Three gaps matter: **Not every AI tool is included.** Google's official definition covers ChatGPT, Gemini, DeepSeek, Copilot, and Grok. Claude and Perplexity are not on that list, so their referrals still land in the Referral bucket, and Google's own AI Overviews and AI Mode get counted as Organic Search rather than as AI traffic. **Most AI traffic has no referrer at all.** This is the big one. An estimated 60 to 70% of real AI-driven sessions arrive with no referrer data, usually because the click came from a mobile app that strips the referrer. ([Loamly's 2026 analysis](https://www.loamly.ai/blog/state-of-ai-traffic-2026-benchmark-report?ref=fallsrivermedia.com) of 446,405 visits put the figure at 70.6%.) Those visits fall into Direct, and the native channel cannot recover them. No tool can fully recover them today. **It does not backfill.** The classification is forward-only. GA4 will not retroactively reclassify the AI traffic you received before the channel went live. If you want any historical view, you have to analyze it separately. The takeaway: the native channel is a convenience, not a complete measurement system. To see Perplexity, to control your own definition of "AI traffic," and to build a record you can trust over time, you still want a custom channel group running alongside it. ## How to Build a Custom AI Channel Group This takes about 15 minutes and gives you a view you control. It runs in parallel with the native channel, so you can compare the two. **Step 1: Open Channel Groups.** In GA4, go to Admin > Data display > Channel groups. You will see Google's default group. Do not edit it. Click "Create new channel group" and name it something clear, like "AI Traffic (2026)." **Step 2: Add an AI channel with a regex rule.** Add a new channel, name it "AI Referral," and set the condition to Source > matches regex, then paste a pattern covering the major sources: ``` chatgpt\.com|chat\.openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|copilot\.microsoft\.com|deepseek\.com|grok\.com|you\.com|meta\.ai ``` Update this quarterly as new tools emerge and as platforms change domains. A single Source condition is all you need here. GA4 combines condition groups with AND rather than OR, so adding a second group such as Medium > exactly matches > `(not set)` would force every session to satisfy that too, which would wrongly exclude AI referrals that arrive with a medium. The Source > matches regex rule on its own already captures your AI traffic regardless of medium, including the no-referrer sessions that show up as `(not set)`. **Step 3: Reorder the channel.** This is the step most guides miss. GA4 assigns each visit to the first matching channel from the top of the list down. If your AI Referral channel sits below Referral, GA4 will file those visits as referrals before your rule ever fires. Drag your AI Referral channel above Referral, then apply and save. **Step 4: View it.** Go to Reports > Acquisition > Traffic Acquisition and switch the channel group selector at the top of the table to your new group. One advantage over the native channel: when you view reports through a custom channel group, GA4 applies its rules to your historical data, so you can see past AI traffic reclassified retroactively, not just from today forward. A couple of technical notes. Free GA4 properties allow only two custom channel groups (GA4 360 allows five), so if you are already at that limit you will need to free a slot first. And since the native AI Assistant channel and your custom rule can both capture the same sessions, read one or the other rather than adding them together. At Falls River Media, we set up AI referral tracking as part of every analytics engagement, native channel plus a custom group for the full picture. ## What to Look For in Your AI Traffic Data Once both views are running, compare AI traffic against your other channels: **Trend line.** The absolute number is small today. Track the growth rate, not just the volume. **Engagement and session duration.** AI-referred visitors are often pre-qualified. They have already read an AI summary about you before clicking. They tend to spend longer on site and view more pages than typical search visitors. **Landing pages.** Which pages are AI tools sending people to? That tells you which of your content AI systems are actually citing, and where your gaps are. **Conversion rate.** Track your key action, whether that is a form fill, a demo request, or a contact. AI referral visitors frequently convert at higher rates than social traffic because they arrive with intent. ## What This Means for Your Strategy If AI tools are sending you traffic, that is a signal your content is being cited in AI-generated answers. That is worth investing in. For B2B companies selling into considered purchases, this channel represents high-intent discovery: someone asked an AI assistant for a recommendation, and you were in the answer. The teams that set up clean measurement now, while volume is small enough to read clearly, will have a year or more of trend data by the time this becomes a meaningful share of traffic. The ones who wait will start from zero. ## Frequently Asked Questions ### Does GA4 track AI traffic automatically now? Partly. Since May 13, 2026, GA4 includes a native AI Assistant channel that automatically separates traffic from assistants like ChatGPT, Gemini, DeepSeek, Copilot, and Grok, with no setup. It does not catch every source (Claude and Perplexity are not on Google's list and still land in Referral, and AI Overviews count as Organic Search), and it cannot recover the majority of AI visits that arrive with no referrer. ### Do I still need a custom channel group? For most businesses, yes, if you want a complete and durable picture. A custom channel group lets you include sources the native channel misses, define AI traffic on your own terms, and run a consistent record you control. The two work well side by side. ### Does setting this up require code changes? No. Both the native channel and the custom channel group live entirely in the GA4 interface. The custom group reclassifies traffic GA4 is already collecting. There is no tracking code to change. ### Can AI tools really see all of my AI traffic? No tool can today. Because so much AI traffic arrives with no referrer, every measurement approach captures a trackable portion and misses the rest. That is still enough to identify trends and measure growth, which is the point. ### Is tracking AI traffic the same as tracking whether AI mentions my brand? No, and it helps to keep them separate. The GA4 setup here measures traffic: the people who clicked through to your site from an AI tool. A different category of tools, AI visibility or AEO platforms, measures whether your brand appears in AI answers in the first place, and they do it by running sample prompts against the models on a schedule, not by watching real conversations, which no one can see. Traffic tracking tells you what happened on your site. Visibility tracking estimates how often you show up in answers. Teams that care about AI discovery usually want both. ### Claude Cowork in 15 Minutes: What to Know, What to Set Up, What to Skip URL: https://fallsrivermedia.com/claude-cowork-in-15-minutes-what-to-know-what-to-set-up-what-to-skip/ Last updated: 2026-07-12T15:22:42.000Z If you have been using Claude in a browser tab and wondered whether Cowork is just the same thing with a different name, the answer is no. The gap is real, and it matters most for people managing multiple clients, ongoing campaigns, and work that lives across a lot of different files and tools. This post covers the essential pieces: what is actually different about Cowork, what to configure in the first 15 minutes, and what you can safely ignore until later. We start with privacy settings, because those should be the first thing you change, not an afterthought. ## What Cowork Actually Is Cowork is a desktop mode for Claude that gives it three capabilities the browser version does not have: **File access.** Claude can read files in a folder on your computer and write deliverables directly back to that same folder. Briefs, decks, reports, and spreadsheets all go to a real location, not a chat window you have to copy from. **Persistent memory.** Claude builds a working knowledge of your clients, projects, preferences, and workflows across sessions. The second conversation picks up where the first one left off without you re-explaining who you are and what you are working on. **Connected tools.** Through plugins, Cowork integrates with your stack: HubSpot, Asana, Gmail, Google Drive, Apollo, and more. You can pull CRM data, check project status, and create or send content without switching applications. These three things compound. A consultant without them has a capable chat assistant. A consultant with them has something closer to a configured team member. ## A Meta-Tip Before You Start: Use Claude Chat to Set Up Cowork If you are already comfortable using Claude in the browser, you have a useful tool for getting Cowork configured: ask Chat to help you. Claude is available in three modes — Chat, Cowork, and Code — and they serve different purposes. Chat is for questions, exploration, and quick tasks that do not require file access or persistent memory. Cowork is for ongoing work with your real files, connected tools, and context that carries between sessions. Code is a command-line tool built for developers. You can use all three, and knowing which one to reach for is part of getting the most out of the platform. The practical tip here: before you configure Cowork, open a Claude Chat session and describe how you work. Tell it about your clients, your tools, how your week is structured, what kinds of tasks take up most of your time. Then ask it directly: which plugins should I install? What should I put in my Cowork memory to start? What connectors are worth setting up first? Chat can reason through your specific situation and give you a setup recommendation tailored to it. That is faster than reading documentation and more useful than a generic starting point. Think of Chat as the place to plan, and Cowork as the place to execute. ## Before Anything Else: Privacy and Data Settings (2 minutes) Consumer plans (Pro and Max) default to allowing Anthropic to use your conversations to improve future Claude models. Most people do not notice this because it is not highlighted during signup. If you are working with client information, campaign data, proprietary strategies, or anything sensitive, you will want to change this before you start. **Where to find it:** Open Claude → Settings → Privacy → "Help Improve Claude" → toggle Off. What that toggle controls: when it is on, new conversations are eligible for model training and are retained for up to five years. When it is off, your conversations are not used for training, and standard 30-day retention applies instead. A few other things worth knowing: **Connector content is excluded either way.** Files Claude reads through connected tools (Google Drive, Asana, HubSpot, MCP servers) are not included in training data regardless of your setting. Only content you paste directly into a conversation window is in scope. **Incognito mode is available for sensitive conversations.** If you have a session involving something you want additional confidence around (a client's financial data, a confidential brief), Incognito mode is never used for training even if you have the model improvement setting turned on. Start a new Incognito chat from the sidebar. **Team and Enterprise plans are already excluded from training by default.** If your organization is on one of those plans, your conversations are not used for model improvement. Still worth verifying with your account admin, but the default is already the protective one. **Plugin and file access: be selective.** Cowork can read files on your computer and take actions in connected apps. Before you grant broad access to a folder, consider what is in it. A dedicated working folder for client projects is better than giving Claude access to your entire desktop or Documents folder. You can also block specific apps from computer use (banking portals, healthcare logins, anything you do not want Claude to encounter) in Settings → Computer Use. Once those settings are confirmed, the rest of the setup takes about 13 minutes. ## The 15-Minute Setup ### Step 1: Connect a folder (2 minutes) When you open a Cowork session, you will be prompted to select a folder. Do not skip this. Choose the folder where your current client work lives. This is where Claude will read context and save deliverables. If you work across multiple clients, start with your broadest working folder, or a single high-priority client to begin. You can change folders between sessions. ### Step 2: Install two plugins (5 minutes) Cowork has a plugin marketplace. You do not need all of it. Start with these two. To get there: click your name in the lower left corner → Settings → Connectors → Customize. You will land on a screen with three tabs: Skills, Connectors, and Personal Plugins. Here is what each one is: **Skills** are individual capabilities Claude can run — write a campaign brief, generate an SEO audit, draft an email sequence. Think of them as individual functions you can invoke on demand. **Connectors** are integrations with specific external tools: HubSpot, Gmail, Asana, Google Drive, and so on. They give Claude access to read and act in those platforms. **Personal Plugins** are pre-packaged bundles that combine relevant skills and connectors together for a specific function. Installing a plugin is the fastest starting point because you get a coherent set of capabilities in one step rather than assembling pieces one at a time. Start in the **Personal Plugins** tab. The two worth installing first: **Marketing plugin.** Covers brand voice enforcement, campaign planning, content drafting, email sequences, competitive research, and SEO audits. One install, most of what a marketing consultant touches week to week. **Productivity plugin.** Covers task management, a memory system, and a daily briefing skill. The daily briefing is the one to try first: ask Claude to brief you on what is open across your projects and it will surface priorities, pending items, and context from your recent sessions. The Skills and Connectors tabs are there if you want to add individual pieces later, but Personal Plugins is the right place to start. Install others (Sales, Operations, Data) once you know you need them. More plugins do not make Cowork faster. ### Step 3: Seed your memory (5 minutes) In your first conversation after setup, describe your work: who your active clients are, what projects you are running, how you prefer to communicate with them, what tools your team uses. Claude will store this and carry it forward. You do not need to write a document or fill out a form. Just talk through your current situation. Something like: "I manage marketing for three clients: a SaaS company focused on mid-market sales, a professional training firm, and a regional real estate group. I primarily work in HubSpot and Asana. I write in a fairly direct, no-jargon style." That is enough to start. Memory updates automatically as you work. You can also tell Claude to remember something specific mid-conversation ("remember that this client prefers monthly reports, not weekly") and it will store that. ### Step 4: Upload your brand guidelines to the brand voice skill (3 minutes) If you manage brand-aligned content for clients, this is the highest-leverage setup in Cowork. Upload a client's brand guide, style guide, or even a document with notes on voice and terminology. After that, every piece of content Claude drafts for that client is checked against those guidelines automatically. For consultants with multiple clients, you can maintain separate brand voice profiles per client. The enforcement is not a post-draft checklist. It runs during generation, which means the first draft is already on-brand rather than requiring a manual review pass. ## Two Features Worth Knowing Now **Scheduled tasks.** You can set Claude to run something on a recurring schedule. A daily briefing at 8am, a weekly pipeline summary on Monday mornings, a monthly report draft on the first of the month. These run at their scheduled time as long as your computer is awake and the Claude desktop app is open — so for a morning briefing to work reliably, the app needs to be running when the time hits. **Skills.** When you ask Cowork to "write a deck" or "build a report," it uses a skill that generates a real file: a .pptx, a .docx, a .xlsx. The output lands in your connected folder. These are not formatted text for you to paste elsewhere. They are working files. ## What to Skip for Now Do not try to connect every tool in your stack on the first day. Start with one connector, the one your work most runs through, whether that is HubSpot, Asana, or Google Drive. Get comfortable with how Cowork handles it before adding more. Do not worry about the data or analytics plugins unless you are actively doing reporting work. They are powerful but add complexity before you need them. The goal in the first week is to run real work through Cowork, not to configure a perfect setup. The memory, the file access, and one or two plugins are enough to see whether and how it fits your workflow. Everything else can follow. **A few caveats.** Cowork and the tools around it are evolving quickly. Some features, including the browser extension, are still in beta, and menu names or settings locations may have shifted by the time you read this, so check what is current in your own account. None of this is legal, compliance, or professional advice. These tools carry real risks, and you are responsible for deciding what client data is appropriate to connect and for reviewing the privacy and security terms that apply to your situation. *Falls River Media works with small teams and marketing consultants to build AI-powered systems that actually fit how they work. If you want help setting up Cowork for your practice, get in touch.* ### What To Do When AI Hands You More Blog Posts Than You Can Use URL: https://fallsrivermedia.com/what-to-do-when-ai-hands-you-more-blog-posts-than-you-can-use/ Last updated: 2026-07-12T15:20:59.000Z A client recently sent me 12 blog posts at once. All generated quickly from a webinar and a couple of live streams. The content was good. The strategy question was harder: what do you actually do with 12 posts? That question is going to land on a lot more desks this year. AI changed the economics of content production almost overnight. A one-hour webinar can now become a dozen drafts in an afternoon. Which is wonderful, until you try to publish, promote, and email all of them and realize your audience isn't actually equipped to absorb 12 posts in a month. The bottleneck moved. It used to be production. Now it's distribution and attention. The editorial calendar has to catch up with that shift, and the way to do it isn't obvious yet. ## The SEO Myth That Made This Worse The instinct to publish all 12 usually comes from an old SEO playbook: more content, more keywords indexed, more organic traffic. That was roughly true a decade ago. It isn't now. Google's recent updates have been aggressive about high-volume publishing that reads as thin or repetitive, especially when posts cluster around the same topics. Twelve posts pulled from one webinar will tend to do exactly that, and the result is keyword cannibalization, where your own posts compete against each other for the same query and split the ranking signal. A stronger move with a content batch is usually one pillar piece that goes deep, plus a few supporting posts that link into it. Topical depth beats topical sprawl. The search algorithm rewards a site that owns a subject, not a site that mentions it a dozen times. There's also a freshness pattern worth knowing. A burst of 12 posts followed by silence reads as abandonment. Steady weekly cadence reads as an active publication. The same 12 posts published over 12 weeks send a much better signal than 12 posts in one week. ## The Email Math Is the Cleanest Argument If a newsletter goes out once a week, and 12 posts hit the blog in a month, three of them get an email push. The other nine sit on the site, found only by people who happen to wander the blog directly. Which is almost no one. You have two reasonable options. Slow the publishing cadence to match the email cadence so every post gets its moment. Or change the email format to a weekly digest that highlights two or three posts, so the backlog still gets surfaced. Both are fine. What doesn't work is publishing on a fast cadence and hoping the posts will find their own readers. They won't. ## LinkedIn Flips the Problem in a Useful Way Social is where the math actually starts to work in your favor, if you stop thinking one blog equals one social post. A single blog can power a quote card, a contrarian take, a short carousel, a clip from the original webinar, and a behind-the-scenes note about why the post got written. That's five or six pieces of social content from one blog, staggered over two or three weeks. Now invert that. Instead of needing to crank out fresh social content every week, you draw from the blog library you already have. The 12-post batch suddenly becomes a quarter of social fuel, released at a sustainable pace, with each post getting multiple chances to be seen. The abundance isn't the problem. The problem is treating each piece as a one-shot announcement rather than a source of weeks of derivative content. ## The Reframe Stop thinking about a webinar as 12 posts. Start thinking about it as one flagship piece plus a library of atomic units you draw from over months. The flagship is the deep, definitive take on the topic. The atomic units are the supporting posts, the social content, the email teasers, the snippets you pull when you need a quick share. Same raw material, very different shelf life. This reframe also fixes the strategy question. You stop asking "how do I get all this content out" and start asking "how do I get the most value from each piece." Those are very different questions, and they lead to very different calendars. ## A Cadence That Holds Up If you want a default to start from, this one works for most small teams: One blog per week. One email per week tied to that blog. Two or three social posts per blog, spread across two weeks. Anything beyond that goes into a backlog, which is not waste. It's runway. Publishing on a steady schedule does something the burst model can't. It gives each post the full chance to land. It signals to search engines that the site is alive. It respects the inbox tolerance of the people who subscribed because they wanted to hear from you, not because they wanted to be buried. The temptation with AI is to treat the output as the goal. The output is just the input to distribution. Production is no longer the hard part. Knowing what your audience can actually absorb is. ### What is Answer Engine Optimization (AEO)? A Practical Guide for B2B Marketers URL: https://fallsrivermedia.com/what-is-answer-engine-optimization-aeo-a-practical-guide-for-b2b-marketers/ Last updated: 2026-06-17T21:41:43.000Z Answer Engine Optimization (AEO) is the practice of structuring your content so that AI tools like ChatGPT, Claude, Perplexity, and Gemini cite it when they answer questions. Where SEO aims to rank your page in a list of links, AEO aims to make your content the source an AI pulls from when it generates a direct answer. The two overlap heavily, but the goal is different. The shift matters because of how buyers now research. A decision-maker evaluating vendors increasingly opens an AI assistant before they open Google. They ask a question, they get a synthesized answer, and that answer either mentions you or it doesn't. AEO is how you influence whether it does. ## AEO, GEO, and the Terminology Question If the acronyms feel like a moving target, you are not imagining it. You will see AEO, GEO (generative engine optimization), LLMO, and AI search optimization used to describe much the same work. The vocabulary debate is largely settled in practice: whatever you call it, the job is to figure out whether AI tools recommend your brand when buyers ask questions in your category, and then to show up more often. There is a useful distinction if you want one. AEO is narrower and answer-specific, about being selected as the direct answer to a question. GEO is broader, about how AI systems describe and recommend your brand across wider conversations. For most B2B marketing teams, you do not need to pick a side. The underlying tactics are nearly identical. One nuance worth knowing: in May 2026, Google published official guidance stating that, from Google Search's perspective, optimizing for its generative AI features is still just SEO. It even named tactics site owners can stop worrying about for Google's own surfaces, including llms.txt files, breaking content into AI-specific chunks, and rewriting pages for every keyword variation. That guidance applies to Google's AI Overviews and AI Mode; other engines such as ChatGPT and Perplexity can weight signals differently. The piece of this that has confused the most people is schema markup, so it gets its own section below. ## AEO vs. SEO: What's Actually Different SEO optimizes for a results page. The win is a high-ranking link a human clicks. AEO optimizes for extraction. The win is being the passage an AI lifts into its answer, often with no click at all. That changes how you measure success. SEO gives you a ranking position. AEO gives you a citation or a mention inside a generated response, which is harder to see and, until recently, harder to measure. The good news is that the foundations overlap: authoritative content, structured data, and topical authority help in both. ## How AI Systems Decide What to Cite AI answer engines favor content that is easy to extract and easy to trust. A few patterns consistently help: **Answer the question first.** Lead with a direct answer in the first 40 to 60 words of the page or section. That is the passage most likely to be extracted. Burying the answer under three paragraphs of setup works against you. **Be specific.** AI systems favor content with numbers, named tools, timeframes, and concrete roles. Specificity reads as authority and gives the model material to cite. **Structure for parsing.** Use question-based headers, short self-contained sections, and scannable formats like bullets, lists, and tables. The cleaner the structure, the easier it is for an AI to map your content to a question. **Build consensus beyond your own site.** AI engines do not only read your website. They weigh your presence across third-party sources: LinkedIn, review sites, industry publications, and other places they look. A strong AEO presence is broader than on-page optimization. ## What About Schema Markup? This is the part of the old AEO advice that has shifted most, and it is worth being precise about, because the common instruction to "add schema to every page so AI can read it" was overstated. Here is Google's actual position, from its May 2026 guidance: structured data is not required for generative AI search, and there is no special schema.org markup you need to add for it. Google's recommendation is to keep using structured data anyway, but for the traditional reason, because it helps you qualify for rich results in regular search. In plain terms, schema is still worth having, just not as your AI strategy. Whatever effect it has on AI visibility is indirect, flowing through better organic performance and clearer signals about who you are, not a direct path to being cited. A concrete example of how much has changed: FAQ schema. For years, adding FAQPage markup to win the expandable question-and-answer boxes under your search listing was standard advice. As of May 7, 2026, Google stopped showing FAQ rich results entirely, finishing a rollback that began in 2023\. HowTo rich results went the same way. The schema types are still valid and Google still parses them, but the visible search feature is gone, so there is no rich-result reason left to add them. So what about ChatGPT, Perplexity, and the other engines? Their crawlers can read schema, and you will find vendors arguing that FAQ and Organization markup are strong AI-citation signals. Be cautious with those claims. There is no confirmed evidence that any major AI engine weights your structured data as a citation factor. These tools work mostly by retrieving and reading your rendered page text, then synthesizing an answer from it. The honest position today is that schema is not a proven AI lever on any platform. What I would actually recommend: - Keep the schema types that still earn rich results and clarify your identity: Organization and Person for entity recognition, plus Article, Product, LocalBusiness, Review, and Event where they apply. This is SEO hygiene, and it is genuinely worth doing. - Do not go hunting for "AI schema" or add FAQ and HowTo markup for visibility. There is no AI-specific schema, and the FAQ rich result no longer exists. Existing FAQ markup can stay; it does no harm. - Put the real effort into the content. What gets you extracted and cited is a clear question answered directly in the visible text, backed by specifics and genuine expertise. A well-written FAQ section still helps for exactly that reason. The value was always in the content, not the markup wrapped around it. The short version: schema is still part of good SEO, but it was never the AI-citation lever it was sold as, and Google has now said so directly. Write for the reader and for clean extraction, and treat schema as hygiene rather than strategy. ## How to Measure AEO Success A year ago, measurement was the weak link. That has changed fast, and there is now a real set of tools. Here is the practical stack, starting with the ones that fit a HubSpot-based marketing operation. **HubSpot AEO.** Launched in HubSpot's Spring 2026 release, this tracks how your brand appears across ChatGPT, Gemini, and Perplexity and gives inline recommendations on what to create or update to improve. Its two genuine differentiators: it suggests prompts based on what HubSpot already knows about your business and buyers from the CRM, and it ties recommendations to specific actions. It is available inside Marketing Hub Pro and Enterprise, or as a standalone product around $50/month. There is also a free AEO Grader for a one-time snapshot of how answer engines currently describe your brand. For teams already on HubSpot, this is the easiest place to start. **Semrush AI Visibility Toolkit.** An add-on to Semrush's SEO suite at roughly $99/month per domain, it tracks how often your brand appears in AI answers across ChatGPT, Google AI Overviews, Perplexity, Claude, and other LLMs, with share of voice, sentiment, and prompt coverage over time. If you already run Semrush for SEO, this folds AI visibility into a workflow you know. **Ahrefs Brand Radar.** A strong benchmarking option built on the largest prompt database in the category, covering Google AI Overviews and AI Mode, ChatGPT, Perplexity, Gemini, and Copilot, with historical data that many newer tools lack. Two things to weigh: it is expensive (priced per AI index, with realistic all-in costs well into the hundreds per month on top of a base Ahrefs plan), and it does not currently track Claude. Best suited to teams already invested in Ahrefs that want directional, competitive AI visibility data. **Manual citation audits.** Free, and still worth doing alongside any tool. Monthly, run 10 to 20 prompts a buyer in your category would actually ask in ChatGPT and Perplexity, and record whether you appear, in what light, and who shows up instead. This gives you direct, unfiltered visibility into how AI represents you. **GA4's AI traffic data.** Pair the above with GA4's native AI Assistant channel and a custom channel group to see the downstream traffic. We cover that setup in a separate guide. ## How These Tools Actually Get Their Data It is worth understanding how these tools collect what they report, because it tells you how much to trust the numbers and where they fall short. Start with what they cannot do. None of these tools can see real user conversations. When someone asks ChatGPT about your category inside their own account, that exchange is private. It is invisible to the AI platforms' own analytics and to every third-party tool. So these tools do not measure what your buyers actually asked. They simulate it. Here is how: **They run a set of prompts on a schedule.** Each tool keeps a list of questions a buyer might ask ("best tools for X," "alternatives to Y," "how does Z compare"), runs them automatically against ChatGPT, Perplexity, Gemini, and the others at regular intervals, and parses each answer for whether your brand appears, in what position, with what sentiment, and which sources got cited. **Where the prompts come from varies, and it matters more than anything else.** Some tools use prompts you enter by hand, some generate synthetic ones, and some build from real search demand. Ahrefs Brand Radar, for example, draws on search-backed prompts rather than invented ones. HubSpot AEO suggests prompts from what it already knows about your business and buyers in the CRM. The closer the prompt set is to what your real buyers ask, the more meaningful the data. A generic or synthetic prompt list can produce confident-looking numbers about questions nobody is actually asking. **They run each prompt multiple times.** LLMs are probabilistic, so the same question can mention your brand on one run and skip it on the next. A single run is noise. Credible tools sample each prompt repeatedly across a measurement window to produce a stable mention rate rather than a coin flip. If a tool reports a number off a single snapshot, be skeptical of it. **They collect the answer one of two ways.** Some tools query the model's API; others capture the answer a real user would see in the product interface. The front-end approach reflects actual exposure more accurately but costs more to operate, which is part of what you pay for at higher tiers. The practical takeaway: treat these numbers as directional, not precise, more like a share-of-voice estimate than a hard metric. The value is in the trend over time and the comparison against competitors, run on a prompt set that reflects how your buyers really ask, not in any single figure. And keep this measurement separate from the traffic tracking in our GA4 guide. These visibility tools estimate whether you appear in AI answers. GA4 tells you who actually clicked through to your site from an AI tool. They answer different questions, and most teams serious about AI discovery end up wanting both. ## Getting Started with AEO If you are already doing solid SEO, creating authoritative content, using structured data, and building topical authority, you are most of the way to AEO. The rest is about structure, specificity, and intent: lead with direct answers, use clean question-based formatting, and optimize for the questions your buyers type into AI tools, not just the keywords they search on Google. For B2B organizations, AEO is not a replacement for SEO. It is the next layer. The companies that start measuring and optimizing for both now will have a real advantage as AI-driven discovery keeps growing. ## Frequently Asked Questions ### Is AEO the same as GEO? Close enough for most purposes. AEO (answer engine optimization), GEO (generative engine optimization), and a few other acronyms describe the same broad goal of getting cited by AI. If you want a distinction: AEO is about being the direct answer to a question, while GEO is about how AI describes and recommends your brand more broadly. ### Is AEO replacing SEO? No. AEO complements SEO. Traditional search still handles enormous query volume, and most website traffic still comes from organic search. Many AEO tactics, like structured data, clear content, and authoritative sourcing, also improve SEO. They reinforce each other. ### How long does AEO take to show results? Usually a few weeks to a few months. Sites with strong existing SEO foundations see faster results because AI systems already recognize their authority. Structural changes like question-based headers and direct answers can influence citations relatively quickly. ### Can small businesses benefit from AEO? Yes, and often disproportionately. AI systems value specificity and authority on a topic, so a focused firm can establish citation-worthy authority in a narrow niche faster than a broad generalist competitor. ### What Your AI Visibility Dashboard Isn't Telling You URL: https://fallsrivermedia.com/what-your-ai-visibility-dashboard-isnt-telling-you/ Last updated: 2026-06-17T21:41:44.000Z There is a number on your AI visibility dashboard that says something like "43% brand visibility." It looks precise. It is not. That number is not a measurement of what AI tools told your buyers. It is an estimate, produced by simulation, and the gap between those two things is the most important thing to understand before you act on any AEO report, hand one to a client, or take one into a board meeting. This is not an argument against the tools. Some of them are very good, and the category is now serious: Profound, one of the larger players, [raised $96 million at a billion-dollar valuation in early 2026](https://www.tryprofound.com/blog/profound-raises-96m-series-c?ref=fallsrivermedia.com). The point is narrower and more useful. If you understand how these tools generate their numbers, you can tell a trustworthy figure from a confident-looking but hollow one, and you can ask the right questions before you spend money or stake a decision on the output. ## The thing nobody puts on the sales page Every AI visibility tool faces the same hard limit: it cannot see what real people ask. When a buyer opens ChatGPT and asks which firms they should consider in your category, that conversation happens inside their private account. It is invisible to the AI platforms' own analytics and to every third-party tool on the market. There is no equivalent of search query data here. Nobody gets to watch the real questions. So the tools do the only thing they can. They simulate. They build a list of prompts a buyer might plausibly ask, run those prompts against the models on a schedule, and read the answers to see whether your brand shows up, where, in what light, and which sources got cited. Everything on the dashboard is built from that simulation. It is a reasonable approach. It is also a model of reality, not reality itself, and it should be read that way. ## The evidence that it's an estimate You do not have to take this on faith. SparkToro ran [one of the more rigorous public studies on AI answer consistency](https://sparktoro.com/blog/new-research-ais-are-highly-inconsistent-when-recommending-brands-or-products-marketers-should-take-care-when-tracking-ai-visibility/?ref=fallsrivermedia.com): 600 volunteers ran 12 prompts through ChatGPT, Claude, and Google's AI a combined 2,961 times in late 2025. The result is striking. There was less than a 1 in 100 chance that any of these tools returned the same list of recommended brands on two runs of the same prompt. Getting the same brands in the same order was closer to 1 in 1,000. Sit with what that means. The same question, asked twice, usually produces two different answers. So the old SEO instinct, "where do we rank," simply does not transfer. You are not in position three for a given query. You appear in some share of responses to it, and that share moves on its own even when nothing about your content or your competitors changes. This is why a single dashboard figure is the wrong unit. "Our mention rate is 43.7%" tells you very little, because you have no stable baseline for what 43.7% means in absolute terms. "Our mention rate on high-intent prompts rose twelve points this quarter" tells you something real. The trend is the signal. The decimal point is decoration. ## The four things that decide whether a score means anything Two tools can report very different numbers for the same brand on the same day, and neither is necessarily lying. The difference usually comes down to four design choices. These are the questions worth asking about any tool, including your own. **1\. Where do the prompts come from?** This matters more than anything else, because a visibility score is only as meaningful as the prompts behind it. Some tools use prompts you type in by hand. Some generate synthetic prompts. Some build from real search demand, which is closer to how buyers actually phrase things. [Ahrefs Brand Radar, for instance, draws on search-backed prompts rather than invented ones](https://ahrefs.com/blog/brand-radar-methodology/?ref=fallsrivermedia.com). HubSpot's AEO tool suggests prompts based on what it already knows about your business and buyers from your CRM. A polished dashboard built on a generic prompt list will give you confident answers to questions nobody is asking. **2\. How many times does it run each prompt?** Given what the SparkToro data shows about variance, a single run is noise. Credible tools run each prompt several times across a measurement window and report a mention rate as a distribution, not a one-off result. As one measurement guide [bluntly puts it](https://graph.digital/guides/ai-visibility/measuring-success?ref=fallsrivermedia.com), running each prompt once per cycle is standard in most off-the-shelf tools, and it is a failure mode: when a single-run number shifts between cycles, you cannot tell a genuine change from normal answer variation. If a tool reports off one snapshot, distrust the number. **3\. How does it collect the answer?** Some tools query the model's API. Others capture the answer a real user would see in the product interface. The front-end approach reflects actual exposure more accurately, but it costs more to operate, which is part of what you are paying for at the higher tiers. API results and front-end results can differ, so two tools using different methods will not match. **4\. How often does it run?** Daily, weekly, monthly. Cadence is partly a budget decision, because every run costs API or capture spend, and partly a data-quality one, because thin sampling produces shaky trends. A sensible pattern is frequent polling for your highest-intent prompts and lighter polling for the long tail. Change any one of these four levers and the headline number changes. That is why cross-tool comparisons are close to meaningless, and why the only fair comparison is a tool against itself over time. ## A short checklist before you trust a number If you are evaluating a tool, or already paying for one, ask the vendor four direct questions: - Which prompts are you running, and where did they come from? - Which engines do you cover, and which do you miss? - How many times do you run each prompt, and over what window? - How do you score a mention, and can I see the underlying answer text? A good tool answers all four plainly and will show you the stored responses behind every score. If a vendor cannot tell you the prompts, the engines, the frequency, and the scoring model, you are [paying for a black box](https://elevatedmarketing.solutions/the-truth-about-ai-visibility-tools-why-they-cant-track-what-they-promise/?ref=fallsrivermedia.com). This skepticism is mainstream now, not contrarian; [marketers across the industry are questioning whether these tools deliver what they promise](https://digiday.com/marketing/marketers-question-expensive-ai-visibility-tools-as-inconsistent-results-fuel-skepticism/?ref=fallsrivermedia.com), and the honest vendors are the ones upfront about where their data has gaps. ## Three things a "visibility score" quietly blurs Even a well-built number hides distinctions that matter for what you do next. **Appearing is not the same as being recommended well.** A brand can show up in nearly every answer and be described as overpriced and unreliable. Zero visibility and bad visibility look identical if all you track is appearance rate. Sentiment and accuracy are separate signals, and in regulated industries an AI confidently misattributing something to your brand is a compliance problem, not a marketing miss. **A mention is not a citation, and a citation is not a click.** Being named in an answer builds association. Being cited as a source can drive a visit. Neither guarantees the other, and most dashboards collapse them. **Visibility is not traffic.** This is the one most teams conflate. AI visibility tools estimate whether you show up in answers. Your analytics tells you who actually clicked through to your site from an AI tool. They answer different questions and they are measured in completely different ways, one by simulation and one by real sessions. We cover the traffic side, and why even that is incomplete, in our guide to tracking AI referral traffic in GA4\. You want both, and you should never read one as a proxy for the other. ## How to actually use the data None of this means the tools are useless. It means they are instruments with a known margin of error, and instruments like that are valuable when you read them correctly. Use them for trends, not absolutes. Watch your mention rate move over quarters, on a prompt set that reflects how your buyers really ask. Use them for competitive gaps, the prompts where a competitor consistently appears and you do not, because those gaps are content briefs in disguise. Use them to catch sentiment and accuracy problems early. And pair the picture with manual spot-checks, because running ten real buyer prompts yourself, by hand, remains one of the clearest windows into how AI describes you. What you should not do is treat a single score as truth, compare two tools' numbers as if they were the same measurement, or report a decimal-point figure to leadership as if it were a fact. The number is a model. Good marketing comes from knowing that, and reading it accordingly. ### Should You Start a Skool Community? URL: https://fallsrivermedia.com/should-you-start-a-skool-community/ Last updated: 2026-06-17T21:41:44.000Z Paid communities are having a moment. One price, one space, courses and a live call and a feed all bundled together, and a room full of people who pay every month to be there. If you sell a service or teach what you know, it is easy to look at that model and feel like it is the obvious next move. It might be. The model is sound, and the pull toward recurring revenue and an owned audience is real. But the platform is the easy part. The hard part is whether you have enough of the right people, and whether you can keep the promise once they pay. Before you pick a name and design a logo, it is worth running the numbers. Here is how to think about whether you are actually ready. ## What you are actually considering For anyone who has not gone down this road yet, Skool is a platform that bundles a discussion feed, a course area, a calendar of live events, and some light game mechanics like points and levels and a leaderboard into one membership. You can run it free or charge for it. The platform itself is inexpensive. There are two plans, nine dollars a month and ninety-nine, with the same features. What differs is the cut Skool takes on member payments, so which plan is cheaper depends on your revenue: below roughly thirteen hundred dollars a month in member payments the Hobby plan usually wins, and above that the Pro plan does. Either way, the platform is not the expensive part. So the cost of the tool is rarely the issue. The cost of your time, and whether anyone will actually show up, is the whole question. ## The promise that makes it worth doing The appeal is real, and it is specific. A service business trades hours for money. Every project starts at zero and ends, and then you go find the next one. A community is the opposite shape. It turns the things you already know how to teach into something you can sell to many people at once, and it keeps paying as long as people stay. It also gives you a standing audience of people who already trust you, which is a much shorter path to a client or a referral than starting cold every time. If you have ever felt the exhaustion of the project treadmill, you already understand the appeal. That part is legitimate. The question is not whether a community is a good model. It is whether you are ready to run one well. ## The part nobody likes to talk about: do you have the audience yet? This is the question most of the hype skips right past. A paid community is not built from strangers. It is built from people who already know you and like what you do. Which means before you think about the platform or the price or the branding, you have to ask an uncomfortable question: how many of those people do you actually have? There are some rough numbers that help you get honest about this. They are not laws, but they are a useful gut check. For a first launch to a warm email list, a common rule of thumb is that somewhere around one to two percent of your subscribers will buy. Not two percent of everyone you have ever emailed, but two percent of the people who actually open and care. So if you want twenty founding members, enough that the community does not feel like an empty room, the simple version of the math says you probably need an engaged email list in the range of one to two thousand people. Want thirty members, and you are looking at closer to fifteen hundred to three thousand. That number stops a lot of people cold, and it should. But it comes with a big asterisk, which is that quality beats size every single time. An engaged, aligned list of a few hundred people who hang on your every email can convert far higher than the one or two percent rule suggests, sometimes ten to twenty five percent. A list of ten thousand people who forgot they signed up will convert close to nothing. The number is a starting point, not a verdict. Now, about YouTube subscribers, or Instagram followers, or any social count. Be careful here, because a big subscriber number feels like proof and usually is not. Social followers convert to paid members at a much lower rate than email subscribers do, because following you is passive and free, while paying you every month is neither. A creator with fifty thousand YouTube subscribers and no email list is often in a weaker position to launch than someone with a fifteen hundred person list of people who reply to their newsletters. The reason is simple: you do not own the YouTube relationship and most of those viewers will never see your offer, while email lands in front of the people most likely to say yes. If you are leaning on a YouTube channel, the honest readiness signal is not the subscriber count on the banner. It is how many of those viewers you have turned into email subscribers, and how many of those actually engage. So the readiness signals worth trusting, in plain terms: - An email list in roughly the low thousands of genuinely engaged people, or a smaller list that is unusually warm and aligned with the exact thing the community would teach. - Evidence that this audience actually responds to you, like real open rates, real replies, people showing up when you go live. - A social following that you have been converting into email subscribers, rather than a big follower count you are hoping to convert later. - Most important, and we will come back to this, proof that they want this specific thing, not just that they like your free content. If you have none of that yet, the answer is not "never." It is "not yet, and here is what to build first." ## The risk to take seriously The thing that should give any owner pause is the empty room. A community is only worth paying for when it is alive. A handful of members in a silent feed is worse than no community at all, because unlike a quiet email list, everyone can see it. Prospects look at a dead community and quietly close the tab. And the moment someone pays, you owe them a live call, an answered feed, and fresh content every week, with no natural finish line the way a project has. Before you open the doors, ask whether you can truly keep that promise on a bad week, not just a good one. If the answer is not a confident yes, the problem is usually timing and readiness, not the idea itself. ## A smarter way to find out You do not have to choose between launching on faith and shelving the idea. There is a path that tests it cheaply first, and it is the one worth following. Start by testing whether anyone wants it before you build a single thing. The cheapest version of this is a waitlist or a short survey to your existing audience asking directly: would you join a paid community about this, and what would you pay. Waitlists, when people opt in specifically, convert far better than a general list, often thirty to fifty percent, so even fifty or a hundred genuine sign-ups can be enough to seed a real founding group. If you cannot get people to raise their hands for free, that is the cheapest and kindest place to learn the answer is no. If the interest is there, open a small founding cohort rather than a permanent open-door community. A founding price, a fixed window of eight to twelve weeks, a clear start and end. That caps what you are committing to, creates urgency, and produces the testimonials and proof you will want before any bigger launch. Pre-build the first few pieces of content before anyone joins, so the doors open onto something real instead of a promise to fill it in later. Choose a live cadence you can keep on your worst week, not your best one. Every two weeks that you never miss beats every week that slips. And decide up front how you will judge it. At the end of the cohort, look at how active people were, whether they finished, whether they want to renew, and honestly, how the time felt against the rest of your work. Then continue, adjust, or stop, based on what actually happened instead of how much effort you have already spent. ## The honest bottom line A Skool community can be a genuinely good move for a small business. The model is sound, and the appeal is real. But the platform is the easy part. The hard part is having enough of the right people, and being able to keep the promise once they pay. If you have an engaged audience in the low thousands, or a smaller one that is unusually warm and pointed at exactly what you would teach, and you can protect the time to show up every week, you may well be ready. If you are mostly counting on a follower number you have not converted into a relationship yet, the most valuable thing you can do is not open a community. It is to spend the next stretch turning that audience into people who would actually raise their hand and pay. Run the numbers before you run the launch. Done well, it does not talk you out of the idea. It tells you whether to do it now, or to do it properly first. --- *Sources:* [*Email list size before launching, Paige Brunton*](https://www.paigebrunton.com/blog/email-list-size-online-course?ref=fallsrivermedia.com)*;* [*Pricing and launching a paid community, CommuniPass*](https://communipass.com/blog/how-to-price-and-launch-a-paid-community-pricing-page-that-converts-even-without-a-big-audience/?ref=fallsrivermedia.com)*;* [*Launch conversion rates, Stephanie Kase*](https://stephaniekase.com/2023/business/bizadvice/typical-conversion-rates/?ref=fallsrivermedia.com)*;* [*Skool pricing 2026, Kourses*](https://kourses.com/skool-pricing/?ref=fallsrivermedia.com) ### Three Years Later: Crafting Content Responsibly in the Age of AI URL: https://fallsrivermedia.com/crafting-content-responsibly-three-years-later/ Last updated: 2026-06-15T16:16:36.000Z In March 2023 I published an article called [Crafting Content Responsibly in the Age of Artificial Intelligence](https://www.linkedin.com/pulse/crafting-content-responsibly-age-artificial-lara-hill/?ref=fallsrivermedia.com). I had been testing ChatGPT for about three months. Midjourney was new to me. Google's Bard had just arrived, and I quoted it in the article partly because quoting an AI about AI ethics felt usefully ironic. Three years later, Bard no longer exists. Neither does the version of the AI conversation we were having in 2023\. This is an honest look back at the four concerns I raised then, what actually happened, and where I land now. The short version: the framework held up better than the details did. ## 1\. Deepfakes, misinformation, and safety **What I said then:** massive amounts of false information could spread at unprecedented scale, our legal systems were not prepared, the EU was about to finalize the first AI law, and the US lacked any regulatory framework. **What happened:** the EU AI Act became law in August 2024, and its main rules begin applying in August 2026, though this spring EU lawmakers reached a provisional deal, still awaiting formal adoption, to push several high-risk obligations out to late 2027\. In the US, the TAKE IT DOWN Act was signed in May 2025, making it a federal crime to publish non-consensual intimate imagery, including AI-generated deepfakes, with platforms required to remove it within 48 hours of notice. As of this spring, 46 states have laws targeting AI-generated synthetic media, and 30 states require disclosure on political deepfakes. **Where I land now:** I was right that the law would lag the technology, but I underestimated how piecemeal the response would be. We did not get one framework. We got a patchwork, and the burden of telling real from fake still falls mostly on the person scrolling. Media literacy turned out to be the durable advice. ## 2\. Privacy, bias, and discrimination **What I said then:** whatever you type into these tools is not confidential, and bias is baked into the datasets because data is a reflection of our past. **What happened:** both points aged well. The dataset problem has not been solved; it has been managed, with mixed results. What changed is that the privacy controls got real. Major tools now offer meaningful settings for data retention and training opt-outs, and I treat checking them as step zero when I adopt anything new. Readers of my [hands-free writing piece](https://fallsrivermedia.com/hands-free-writing/) may remember that the first thing I did with my dictation app was turn off data retention. **Where I land now:** the questions I borrowed from Coded Bias and Cathy O'Neil are still the right ones. Who does this system decide against, and would they ever know? For small businesses using AI in hiring, lending, or customer decisions, this stopped being theoretical: state regulators are now watching automated decision tools closely. ## 3\. Transparency **What I said then:** OpenAI had gone closed, and I shared Cathy O'Neil's warning that AI is really about power, because it is all about who owns the code. **What happened:** the most surprising reversal of the three years. Open-weight models caught up to the frontier. DeepSeek's releases matched leading proprietary systems on major benchmarks, and OpenAI itself released open-weight models under a permissive license, something that felt unimaginable when I wrote the original. Meanwhile California passed SB 53, the first US law requiring frontier AI developers to publish their safety frameworks, effective January 2026. **Where I land now:** the open-versus-closed question turned out to be less settled than it looked in 2023, and that is good news. But O'Neil's deeper point about asymmetrical power has not budged. A handful of companies still decide what these systems optimize for. Transparency laws help. They do not balance the scales. ## 4\. Copyright and legal **What I said then:** if you use AI to generate content, you do not own it, and anyone can take it without legal ramifications. **What happened:** this is the section I most need to correct. The picture is now far more nuanced. The US Copyright Office's 2025 guidance confirmed that purely AI-generated work is not copyrightable, and that prompts alone, however detailed, do not make you an author. But it also confirmed that works combining human and AI contributions are assessed case by case, and the human contributions can absolutely be protected. The training side exploded into the courts: a federal judge found that training on legally purchased books was fair use while holding that keeping pirated copies was not, which led to a $1.5 billion settlement, the largest in US copyright history, now awaiting final court approval. The New York Times case against OpenAI is still moving. European courts have begun ruling against AI developers on training data. **Where I land now:** for working content creators, the practical rule is to document your human contribution. If AI drafted it and you shaped it, edited it, structured it, and made the creative calls, you have a much stronger position than 2023 me believed possible. Total protection and zero protection were both wrong answers. As before, I am not a lawyer and this is not legal advice. But the direction of travel is clear: human authorship is the asset. Protect it by actually doing it. ## What I would tell a small team in 2026 The 2023 article ended by saying we would inevitably experience the negative impacts along with the positive unless stakeholders built governance and individuals educated themselves. That held up. The governance is arriving slowly and unevenly, which means the self-education part is still carrying most of the weight. Here is what that looks like in practice for the small teams I work with: 1. **Check the data settings before you adopt any tool.** Retention, training opt-outs, and where your client information goes. Five minutes, once, per tool. 2. **Keep a human visibly in the work.** Not as a compliance gesture, but because human authorship is now both your legal position and your differentiator. 3. **Disclose in ways that build trust.** Your audience assumes AI is involved somewhere. Telling them how you use it, the way I did in my hands-free writing piece, reads as confidence rather than confession. 4. **Revisit your assumptions yearly.** Most of what I believed in March 2023 needed updating by 2026\. Whatever you believe today has the same shelf life. The wave I wrote about three years ago did not crash and recede. It became the water we work in. I am still convinced the right response is neither refusal nor surrender, but literacy: knowing what these tools do, what they cost, and who they affect. That was the point in 2023\. It is still the point. If this sparked something, I would love to hear it. [Find me on LinkedIn](https://www.linkedin.com/in/larahill?ref=fallsrivermedia.com). --- ### Sources - EU AI Act [implementation timeline](https://artificialintelligenceact.eu/implementation-timeline/?ref=fallsrivermedia.com) and the May 2026 [Digital Omnibus amendments](https://www.insideprivacy.com/artificial-intelligence/eu-ai-act-update-timeline-relief-targeted-simplification-and-new-prohibitions/?ref=fallsrivermedia.com) - [TAKE IT DOWN Act](https://en.wikipedia.org/wiki/TAKE%5FIT%5FDOWN%5FAct?ref=fallsrivermedia.com) and MultiState's [state deepfake law tracker](https://www.multistate.us/insider/2026/2/12/how-ai-generated-content-laws-are-changing-across-the-country?ref=fallsrivermedia.com) - US Copyright Office, [Copyright and Artificial Intelligence](https://www.copyright.gov/ai/?ref=fallsrivermedia.com) reports (Parts 1 to 3) - Kluwer Copyright Blog on the [Bartz v. Anthropic settlement](https://legalblogs.wolterskluwer.com/copyright-blog/the-bartz-v-anthropic-settlement-understanding-americas-largest-copyright-settlement/?ref=fallsrivermedia.com) and Norton Rose Fulbright's [2026 AI copyright case update](https://www.nortonrosefulbright.com/en/knowledge/publications/ce8eaa5f/ai-in-litigation-series-an-update-on-ai-copyright-cases-in-2026?ref=fallsrivermedia.com) - Future of Privacy Forum, [California SB 53 explained](https://fpf.org/blog/californias-sb-53-the-first-frontier-ai-law-explained/?ref=fallsrivermedia.com) - My original 2023 article on [LinkedIn](https://www.linkedin.com/pulse/crafting-content-responsibly-age-artificial-lara-hill/?ref=fallsrivermedia.com) *Facts checked as of June 2026.* ### Is Your Lovable Site Findable by ChatGPT and Claude? URL: https://fallsrivermedia.com/is-your-lovable-site-findable-by-chatgpt-and-claude/ Last updated: 2026-06-15T16:16:49.000Z ## What the May 2026 update means for anyone building or buying a B2B website right now If you built your website on Lovable in the last year, there's a good chance your site was invisible to AI assistants like ChatGPT, Claude, and Perplexity. That's not a guess. It was a known technical limitation of how the platform rendered pages. As of May 13, 2026, that's no longer true. Lovable shipped an update that addresses the problem for new and existing projects. Here's what changed, what you need to do if you built before the update, and what the whole story tells us about evaluating any AI website builder for marketing. ## The old problem Until mid-May, every site built on Lovable was a React single-page application with client-side rendering. The short version of why this matters: when a search engine or AI crawler visited the site, it received a nearly empty HTML shell. The actual content (your headlines, your value proposition, your service descriptions) was generated by JavaScript after the page loaded in a browser. A human visitor didn't notice. A bot often saw a blank page. This wasn't a Lovable-specific bug. It's the default behavior of any platform built on modern JavaScript frameworks without server-side rendering. But it had real consequences for marketing sites. Google indexed pages slowly or not at all. AI crawlers, which mostly don't run JavaScript, saw nothing. For B2B companies hoping for organic search traffic or AI citations, this was a meaningful problem. ## What Lovable shipped on May 13 According to [Lovable's own announcement](https://lovable.dev/blog/building-apps-using-tanstack-start?ref=fallsrivermedia.com), new projects created on or after May 13, 2026 are now built on a framework called TanStack Start with server-side rendering enabled by default. Every page request returns fully rendered HTML to both humans and crawlers. For older projects, Lovable shipped a different fix. [Per Lovable's documentation](https://docs.lovable.dev/features/seo-aeo?ref=fallsrivermedia.com), legacy React and Vite projects now use on-request pre-rendering on deployed public URLs. When a verified crawler arrives (Google, Bing, social-preview bots, and AI engines including ChatGPT, Perplexity, Claude, and Gemini), Lovable renders the page on the fly and returns the resulting HTML. The release also bundled in an SEO and AI search review tool, Semrush integration in chat, and one-click fixes for common metadata issues. [SEO consultant Till Freitag's analysis](https://till-freitag.com/en/blog/lovable-seo-aeo-discoverability-en?ref=fallsrivermedia.com) called it Lovable shifting from app builder to app distribution layer. In plain terms: the rendering problem is largely solved, both for new projects and for older ones that previously had it. ## If you built on Lovable before May 13 If your project predates the update, you don't get full SSR. You get automatic pre-rendering instead, and there are a few things worth knowing. **The pre-rendering is automatic.** You don't have to enable, configure, or migrate anything. Per Lovable's documentation, static HTML snapshots are generated automatically when verified crawlers (Google, Bing, ChatGPT, Claude, Perplexity, Gemini, and social-preview bots) request a page on your deployed public URL. You also don't need to pay for or install any third-party prerendering service. **But you can't verify that by looking at your site.** Because pre-rendering only serves to verified crawlers, the view-source experience in your browser will still look like a single-page app. To see what crawlers actually see, you need to check from a crawler's perspective. Here's how I'd approach it. **Use Google Search Console first.** This is an important (and free!) AEO and SEO diagnostic tool, regardless of where your site is hosted. If you haven't already, verify your domain in Search Console and submit your sitemap. Then use the URL Inspection tool to view your live URLs and confirm Google is seeing the rendered content (headlines, body copy, meta tags, structured data). Search Console will also surface indexing issues, coverage problems, and which queries are surfacing your site. For most websites, Search Console is the closest thing to ground truth on whether you're actually findable. **Cross-check with Semrush or another SEO platform.** Search Console tells you what Google sees. Semrush (or Ahrefs, or similar) tells you how your site is performing competitively: which keywords you're ranking for, what your competitors are ranking for that you aren't, and where you have backlink and content gaps. For AEO specifically, Semrush has added features that track AI assistant visibility, including whether your domain is being cited in AI Overviews and similar AI-generated answers. This is where the strategic AEO conversation actually happens, not in the platform's built-in tools. **Use link debuggers to verify social previews.** LinkedIn Post Inspector, Facebook Sharing Debugger, and similar tools show you the metadata that bots actually receive when your link is shared. These are quick, free, and useful for confirming that Open Graph tags, page titles, and preview images are coming through correctly. **Then decide whether to migrate or stay.** This is the real strategic question for legacy users. Pre-rendering is good. SSR is better. The difference matters most if you're publishing a lot of new content, or if you're noticing that dynamically loaded content (think filtered listings, search results, or anything that updates after page load) is not getting indexed. For a static marketing site that publishes occasionally, pre-rendering is likely enough. For a content-heavy site or one where AEO is core to your strategy, rebuilding on TanStack Start is worth considering. There is no automatic migration path from legacy projects to the new stack; you would need to rebuild. If you're not sure which side of that line you're on, your Search Console data plus a Semrush check of your AI visibility is usually enough to tell you. If the indexing is clean and you're being cited where it matters, stay. If you're invisible in AI answers despite good content, the migration starts to make sense. ## What this changes If you've been considering Lovable for a marketing site, the conversation is different now than it was a month ago. Specifically: - **New Lovable projects ship with real SSR by default.** No workarounds, no prerendering services, no migration. The thing your buyers' AI assistants need to see your content is in the response from the first request. - **Existing Lovable projects get automatic pre-rendering for verified crawlers.** This is not as clean as native SSR (third-party SEO scanners and unverified bots still see the SPA shell), but it solves the practical visibility problem for Google and the major AI engines. - **Setup-side AEO basics are getting easier.** Meta tags, Open Graph data, and structured data are flagged in the review tool and fixable in one click. There are still caveats. [Per Lovable's FAQ](https://lovable.dev/seo-aeo?ref=fallsrivermedia.com), full SSR is currently only available for new projects on TanStack Start, not retroactively for legacy projects. And the technical foundation being solid is not the same thing as having a good content strategy, the right schema markup for your industry, or content that AI engines actually want to cite. But the central problem with Lovable's AEO foundation is, for most practical purposes, fixed. ## Why this matters beyond Lovable The pattern here is what's interesting, not just the specific platform. Three things are worth taking from this: **One: the underlying concept (server-side rendering vs. client-side rendering) is now a question you should be asking about every platform you evaluate.** Not every AI website builder defaults to SSR. Some still serve empty HTML shells to AI crawlers. The question isn't whether the platform is fashionable. It's whether your content reaches the bots that decide whether you exist in an AI-generated answer. **Two: AI-built tools are evolving faster than most evaluation cycles.** A piece accurate three weeks ago can be outdated today. That happens now. If you're doing platform evaluations on a quarterly or annual cadence, you're probably making decisions on stale information. This is one of the real second-order effects of the current AI tooling pace, and it deserves more attention in B2B marketing operations than it currently gets. **Three: "fixed by default" is not the same as "configured well."** Even with Lovable's update, the platform doesn't automatically set up canonical tags, meta descriptions, structured data, llms.txt, or content optimized for AI citation. Those still take real work. The rendering problem is solved. The strategic work isn't. ## What to ask before you commit to any AI website builder If you're choosing a platform right now (Lovable or anything else), the diligence questions worth asking have shifted. They're now: 1. Does the platform serve fully rendered HTML to crawlers by default, or only with additional setup? 2. If my project is older, does the platform pre-render for AI crawlers automatically, or do I need a third-party service? 3. What does the platform handle automatically (metadata, schema, sitemaps, llms.txt) and what's left to me? 4. How do I verify what AI crawlers actually see when they visit my site? 5. What's the migration path if I outgrow the platform or it changes direction? These are the questions that matter for AEO. The platform marketing won't always answer them clearly. Lovable's recent update is a good example of a platform getting this right. Other platforms in the same category still haven't. ## The bigger pattern The job of a marketing leader hasn't changed. It's still to ask the annoying questions before the contract gets signed, and to keep asking them after, because the answers change. AI-built websites can absolutely be made to work for AEO. Lovable's update is proof. But "built by AI" and "findable by AI" are still not automatically the same thing, and platforms move at different speeds. Knowing the difference, and knowing what to ask, is the work. That's where smart B2B marketing teams have an edge right now. Not by picking the trendiest tool, and not by avoiding new platforms entirely, but by understanding the technical fundamentals well enough to evaluate any of them on the things that actually matter. --- ### Sources - Lovable, [Building apps using TanStack Start](https://lovable.dev/blog/building-apps-using-tanstack-start?ref=fallsrivermedia.com) - Lovable, [SEO and AEO documentation](https://docs.lovable.dev/features/seo-aeo?ref=fallsrivermedia.com) - Lovable, [SEO and AI search product page](https://lovable.dev/seo-aeo?ref=fallsrivermedia.com) - Till Freitag, [Lovable SEO/AEO Release: SSR, Pre-Rendering and More](https://till-freitag.com/en/blog/lovable-seo-aeo-discoverability-en?ref=fallsrivermedia.com) - Rasesh Koirala, [SEO For Lovable Websites (Updated April 2026)](https://raseshkoirala.com/blog/seo-for-lovable-websites/?ref=fallsrivermedia.com) - Hado SEO, [Lovable SSR and TanStack Start: What It Means for SEO](https://hadoseo.com/blog/lovable-ssr-update-tanstack-start-seo?ref=fallsrivermedia.com) --- *Falls River Media helps small teams and solo experts build AI-augmented marketing systems that are designed to be found by search engines and AI assistants. If you're rethinking your website or wondering whether your current setup is AEO-ready, we'd love to talk.* ### What Finally Made Hands-Free Writing Work for Me URL: https://fallsrivermedia.com/hands-free-writing/ Last updated: 2026-06-15T16:16:50.000Z Two years ago I wrote an article about creating content hands-free while picking blueberries in my garden. The response told me something: a lot of people are trying to figure out how to work smarter, and apparently the image of someone dictating a LinkedIn article while harvesting fruit in their backyard resonated. At that time I still wasn't fully happy with the workflow. It worked, but it had friction. Every tool was separate. Every step required switching apps. The voice dictation tools required correcting way too often. The blueberries are flowering right now. The harvest is a few months away. And I'm already planning to be back out there this summer, phone by my side, but using a workflow with less friction. If you have ever wanted to write more but couldn't find the time to sit down and type at a keyboard, this article is for you. I'm going to walk you through exactly how I wrote this piece using only my voice and two AI tools. By the end, you'll have a workflow you can try today for writing something you're proud to publish, no typing required for most of it. ## Introducing Wispr Flow\* The tool that has changed everything for me lately is [Wispr Flow](https://wisprflow.ai/r?LARA334&ref=fallsrivermedia.com)\*, a voice dictation app that works across every app on every device I use. Texting, email, AI prompting, it doesn't matter. If I can type in it, I can dictate into it with Wispr Flow. What sets it apart from every other dictation tool I've tried, including the built-in iPhone voice dictation I used to struggle with, is its ability to capture what I intended to say rather than what I literally said. When I misspeak, it doesn't transcribe the mistake. It figures out what I meant and writes that instead. I've been using it for about two weeks. In that time I've voice-dictated over 28,000 words that I never had to type. I didn't fully realize until then how many words I type in everyday communications alone. We tend to think of 28,000 words as something a novelist produces in a couple weeks of serious writing. But that's just my regular work life: emails, texts, prompts, messages. Part of why the volume adds up so fast is speed. Wispr Flow clocks my average dictation at 122 words per minute. The average person types somewhere between 40 and 60 words per minute. That gap is significant when you're trying to capture ideas quickly, keep up with your own thinking, or simply get through your inbox faster. It has made me twice as fast! Before I committed to using it daily, I checked the privacy settings carefully. Wispr Flow allows you to opt out of data retention and model training entirely. I updated my settings so that none of my voice-dictated data is retained, used for training, or kept past 24 hours. For anyone working with client information or simply uncomfortable with voice data being stored, that's worth knowing and worth doing before you start. ## How I wrote this article, step by step Here is the exact workflow I used to write this piece. I'm sharing it this specifically because I want you to be able to try it yourself. 1. **Brain dump by voice.** I started by opening Claude on my phone and dictating my initial thoughts using Wispr Flow. Not polished sentences. Just ideas, memories from the original article, things I wanted to say, questions I wanted to answer. Think of this as a conversation with a very patient thinking partner who captures everything you say. 2. **Ask Claude to find the structure.** Once I had my brain dump in the chat, I worked with Claude to find the structure, key themes, and clarify what I was actually trying to say. 3. **Draft by voice.** Using Wispr Flow, I dictated revisions. "Change the third paragraph to say this instead." "Make the opening shorter." 4. **Ask Claude to email the draft.** Once I was happy with the draft, I asked Claude to put it in a format that I could easily email to myself. 5. **Later at my desktop.** I copied the draft from my email to my article. I made some final edits. This was the one step that required a keyboard. That's it. Five steps, mostly by voice, from a scattered brain dump to a publishable article. ## Tools mentioned in this article [Wispr Flow](https://wisprflow.ai/r?LARA334&ref=fallsrivermedia.com)\*: voice dictation that works across every app on every device. [Claude](https://claude.ai/?ref=fallsrivermedia.com): I'm guessing you've heard of this one lol. ## Closing I started writing about this topic two years ago in my garden because I genuinely needed a better way to work. I'm a working parent with multiple businesses, a full schedule with lots of travel, and a deep belief that how we work matters as much as what we produce. Spending hours hunched over a keyboard when you could be outside, moving, living, has always felt like the wrong trade to me. The blueberries are flowering. The harvest is coming. And for the first time, I feel like the tools are actually ready to meet me where I am. If you try this workflow, I'd love to hear how it goes. [Join the conversation on LinkedIn](https://www.linkedin.com/pulse/what-finally-made-hands-free-writing-work-me-lara-hill-zjajf/?ref=fallsrivermedia.com). *\*Wispr Flow referral: use my link and we both get a free month of Pro, which includes unlimited transcription.* --- *Originally published on* [*LinkedIn*](https://www.linkedin.com/pulse/what-finally-made-hands-free-writing-work-me-lara-hill-zjajf/?ref=fallsrivermedia.com)*, April 6, 2026.*