A new line item showed up in your 2026 marketing budget. It’s called AEO. Or GEO. Or AI Search Optimization, depending on which vendor’s deck landed in your inbox this week.

You’re being sold the same lie SEO sold you in 2014.

The pitch goes like this: ChatGPT and Perplexity are eating your organic traffic, so you need a new optimization layer. Schema markup. FAQ blocks. “LLM-friendly” content structures. Pay a consultant eight thousand a month to tell you to add bullet lists. Scrunch’s April 2026 roundup names seven “leading” AEO tools. Adobe shipped an LLM Optimizer. Semrush has an AI Visibility Toolkit. Every agency that lost SEO retainers in 2024 has rebranded as “Generative Engine Optimization.” Same retainer. New acronym. Same client.

There’s one problem nobody pitching you wants to talk about.

AI engines aren’t really reading your website. 🤖

Nobori analyzed more than 200,000 AI prompt runs tied to commercial B2B discovery and found that 85% of AI search visibility comes from off-site sources. Not your beautifully schema-tagged landing pages. Not your “ultimate guide” pillar content. Reddit threads. G2 reviews. Podcast transcripts. Substack posts. The Hacker News comment your founder wrote in 2022. That’s the corpus ChatGPT is summarizing when somebody asks “what’s the best vendor for [your category].”

You can rewrite every H2 on your site in markdown bullet syntax. It will not move the number that matters.

The trend underneath that number is not in dispute. Cognism’s Inside Inbound 2026 report — and they were honest enough to publish their own GA4 data — showed organic sessions down 33.6% year over year and clicks down 23% in the last six months. Meanwhile direct MQLs grew 6%. That gap is the entire story. Buyers are still buying. They’re just doing the research somewhere you can’t measure and arriving on your site after they’ve already shortlisted you.

Wynter’s January 2026 survey of mid-market SaaS CMOs said the same thing in cleaner language: 84% of CMOs now use AI tools like ChatGPT, Claude, and Perplexity for vendor discovery, and 68% start their search in an AI assistant before they ever type into Google. These are the people buying your software. They’re not reading your blog. They’re asking an LLM “who should I look at for [problem]” and accepting a three-vendor shortlist based on what other people have written about those vendors.

So when your AEO consultant tells you the answer is more on-site optimization, what they’re actually selling is the comforting illusion that you can solve this with a CMS change. 🤡

You can’t. The corpus has moved.

Here’s the tell that AEO is mostly cosplay. Walk through any of the new “GEO” playbooks and count how many of the recommendations are things SEO consultants were already selling in 2021: cleaner schema, FAQ blocks, structured author bios, internal linking, “answer-first” content. The packaging changed. The artifact didn’t. Nobori’s own broader dataset puts B2B AI visibility tracking adoption at 47% — up from 8% a year earlier — but most of those teams are tracking visibility without doing the off-site work that would actually move it. They bought the dashboard. They didn’t change the strategy.

Yesterday’s recap from B2BMX 2026 in Demand Gen Report captured the same idea from the buyer side. LinkedIn’s Ty Heath introduced what she calls “buyability” — the argument that deals don’t die because competitors win. They die because the buying group lacks collective confidence. Her work with Bain found that 81% of won deals had broad awareness across the buying committee from day one. When only the champion knew the vendor? The deal stalled.

Buyability and off-site visibility are the same problem in two languages. Your champion needs five other people on the buying committee to say “yeah, I’ve heard of them.” Where do those five people form that opinion? Not on a page you control. They form it on Reddit, in a private Slack community, on the podcast they listened to on the way to the office, in the G2 thread their procurement person forwarded around. If you’re not in those places, your champion walks into the deal review alone, and the deal is dead before the demo.

This is what the AEO vendors are technically right about and tactically wrong about. AI visibility matters more than ever. They just keep pointing it at the wrong fix.

Run it like a PR project, not an SEO project. Here’s the actual playbook.

First, do the audit that matters: ask the LLMs themselves. Open ChatGPT, Perplexity, Gemini, Claude. Type the buyer’s actual question — “best vendor for [category] for a Series B company” — and read what comes back. Read the citations. Now you know the corpus that’s deciding your category. If you’re not in it, you’ve found your problem. If you are in it but the description is wrong, you’ve found a different problem. Schema tags fix neither.

Second, look at the citation sources. Almost every B2B category surfaces the same shape: a couple of analyst pages, one or two Reddit threads, G2 or a comparable review site, sometimes a YouTube video, occasionally a podcast or a Substack. That list is your distribution plan. It is not a content calendar. It is a list of relationships, threads, and shows you actually need to be in.

Third, treat customer reviews like a real marketing channel, not a quarterly chore. G2 still moves the needle, and the current generation of LLMs scrapes it aggressively. Twenty fresh reviews with specific use-case language will do more for your AI visibility this quarter than three months of on-site “thought leadership.” The AEO vendors know this. They just can’t bill against it.

Fourth, get your customers and your subject-matter experts into the off-site corpus under their own names. Not branded sponsored content. Actual posts. Actual podcasts. Actual Reddit replies. The Hacker News comment your founder wrote in 2022 is in the training data right now. Your sanitized fifteen-page “definitive guide” PDF isn’t, because nobody linked to it and Reddit didn’t argue about it. 📉

Fifth — and this is the unfashionable one — fund brand. Mental availability isn’t a 2018 D2C concept that doesn’t apply to B2B. Heath’s buyability research is the same argument with a B2B label slapped on. Buyers buy from vendors their committee has already heard of. The LinkedIn–Amazon Connected TV deal announced on May 8, which puts LinkedIn’s B2B audience graph inside Amazon DSP, is the same trend at a different price point. The smart money is moving back into being known, because being found doesn’t work the way it used to.

Compare two B2B companies in the same category. Company A spent Q1 stuffing their site with “What is [category]” answer blocks and bragging on LinkedIn about their GEO score. Company B spent Q1 getting four customers to write detailed G2 reviews, sending two engineers onto a niche industry podcast, and seeding three Reddit threads with their head of product answering questions under his own name. Ask ChatGPT next week which company gets cited. It will not be the one with the prettier FAQ schema. It will be the one whose name is already in the conversation.

The dumb money is still going into AEO retainers and another “AI SDR” tool. GTM-AI funding has cleared $2.7 billion in 2026 already, per 365Outsource’s roundup of recent rounds. Most of it is being spent to automate the activity layer of a funnel that buyers no longer walk through.

Google Marketing Live is on Tuesday. The agentic-ads narrative is about to flood your LinkedIn feed for forty-eight hours. Every vendor will have an “AI campaigns” announcement to retrofit into Wednesday’s webinar slides. Ignore most of it. The single most useful thing you can do this quarter is run the LLM audit, find the off-site sources actually being cited in your category, and reallocate at least 20% of your content budget from on-site posts into off-site placement, review velocity, podcast appearances, and named expert participation.

There’s a clean test for whether your team has an AI search strategy or just an AEO invoice. Ask whoever owns the line item one question: “What does ChatGPT say, today, when somebody types our category and asks for a vendor recommendation?”

If they can’t answer in under sixty seconds, you don’t have a strategy. You have a vendor. ✋