Last month I wrote about the pipeline paradox — more activity, less pipeline — and the quiet collapse of top-of-funnel search traffic. Several readers asked the obvious follow-up: if buyers aren't finding us through Google anymore, where are they, and how do we get in front of them?
Here's the uncomfortable answer. Before your ideal buyer ever lands on your website, fills in a form, or replies to an email, they've already had a conversation about your category — with an AI. And in that conversation, a shortlist of vendors got named. You were either on it or you weren't.
This is the single biggest shift in B2B buying I've seen in fifteen years, and most companies I talk to still have no strategy for it.
The buying journey now starts inside a chatbot
The numbers moved faster than anyone expected. Forrester's 2026 buyers' survey of nearly 18,000 business buyers found that 94% now use generative AI somewhere in their purchase process, up from 89% a year earlier — and that buyers named conversational AI as their most meaningful research source, ahead of vendor websites, analysts, and sales reps.
G2's 2026 research puts a finer point on it: 51% of B2B software buyers now begin their research with an AI chatbot more often than with Google, up from 29% in April 2025. Roughly seven in ten lean on AI to compare vendors before they ever talk to one.
But the statistic that should change how you spend your time is this one: 69% of buyers chose a different vendor than they originally intended based on what the AI told them, and one in three bought from a vendor they'd never heard of before the AI named it.
Read that again. A third of deals are now going to companies that weren't on the buyer's radar at all — they were introduced by the model. That cuts both ways. If your competitor is the one the AI recommends, your brand recognition, your case studies, and your relationships may never get a chance to matter. If you're the one it recommends, you can win accounts that have never heard of you.
Why your SEO content isn't doing this for you
The instinct is to assume that if you ranked well on Google, you'll show up in AI answers too. You won't, not reliably. Answer engines like ChatGPT, Perplexity, Gemini, and Claude don't work like a search results page. They don't return ten links and let the buyer choose. They synthesise a single recommendation — and they assemble it from a different set of signals than Google's algorithm ever used.
Three things matter to a model that barely mattered to classic SEO:
Entity clarity
The model needs to understand what you are with zero ambiguity — your category, who you serve, the specific problems you solve, the regions you cover. Vague positioning ("we help businesses grow") is invisible to a system trying to match a buyer's precise query to a precise vendor. Sharp positioning is now a discoverability advantage, not just a messaging one.
Third-party corroboration
Models trust consensus across sources far more than they trust your own website. Being described consistently — on review platforms, in directories, in podcasts, in other people's articles, in community threads — is what gets you named. Your homepage saying you're great counts for almost nothing. Twenty other places independently confirming what you do counts for a lot.
Structured, quotable answers
Content written as clear questions and direct answers gets retrieved and cited. Meandering thought-leadership and 2,000-word "ultimate guides" mostly don't. The model wants the clean, liftable sentence.
What actually works: a practical playbook
I've been running this with clients across SaaS, professional services, and insurance tech over the last two quarters. Here's what's moving the needle.
1. Find out what the AI says about you today
Start with the audit nobody does. Open ChatGPT, Perplexity, Gemini, and Claude, and ask the questions your buyers actually ask: "Best [your category] for [your buyer] in [your region]?" "Alternatives to [your biggest competitor]?" "Who should a mid-market logistics firm in the DACH region talk to about X?" Write down who gets named, in what order, and how you're described. This is your real ranking now — and it's usually a humbling read.
2. Fix your entity, not just your copy
Make it impossible to misunderstand who you are and who you're for. Tighten the positioning on your site, then make sure the same crisp description appears everywhere the model reads — your LinkedIn company page, your directory and review listings, your founder bios. Consistency across sources is the signal. Contradiction is the killer.
3. Build third-party presence deliberately
Because corroboration is what earns the citation, the highest-leverage work is often off your own website. Get listed and reviewed on the platforms your category trusts. Get quoted in other people's content. Show up on the podcasts and in the communities where your buyers — and the models scraping them — are paying attention. In Europe especially, don't ignore the regional and industry-specific directories; they carry disproportionate weight for localised queries.
4. Publish to be quoted, not just to rank
Restructure your best content around the literal questions buyers ask, with the answer in the first two sentences and the nuance after. Lead with a clear, liftable claim. Include the specifics models love — numbers, named use cases, comparisons, who it's not for. The April rule still holds: if your competitor could have published the same post, it won't get cited. A genuine point of view is now a retrieval advantage.
5. Make your social proof machine-readable
A third of buyers are buying from vendors they didn't know — on the strength of what the AI surfaces. If your case studies, results, and customer language live only inside a gated PDF or a sales deck, the model can't see them. Get your proof onto indexable, public pages where it can be read, matched, and quoted.
The new top of funnel
Here's the mindset shift I'd leave you with. For a decade we optimised to be found — to rank, to capture the click, to win the visit. The job now is to be recommended — to be the name the model offers up when your buyer asks for help, before any visit happens at all.
The old question was "How do we rank for this keyword?" The new question is "When our buyer asks an AI for the best option, are we the answer it gives — and can we prove we deserve to be?"
This isn't a reason to abandon everything that works. Your website, your content, your reviews, and your reputation are exactly the raw material these models draw on. But they're being read by a new kind of reader — one that summarises instead of listing, recommends instead of ranking, and reaches your buyer long before you do. The companies winning the AI shortlist in 2026 aren't the ones shouting loudest. They're the ones who are clear about what they are, corroborated everywhere it counts, and impossible for the machine to leave out.
Wondering what the AI says about you?
Let's run the shortlist audit on your category — and build a plan to make your company the name the model recommends.
Book a free strategy call →