How B2B Buyers Use AI Search
B2B buyers now ask ChatGPT and Gemini to build a vendor shortlist before Google, drawing on indexed pages and third-party reviews rather than your homepage.
B2B buyers increasingly open an AI assistant before a search engine, asking it to name vendors, summarise pricing or explain a category, and use the answer as a first shortlist. Assistants build that answer from indexed pages and third-party sources, not from a vendor's own claims about itself.
That shift is the reason B2B SEO now treats assistant visibility as part of the same programme as ranking, rather than a separate project.
How does a B2B buyer actually use an assistant when evaluating vendors?
Usually early and briefly. A buyer asks a broad question — "what are the best tools for X" or "how does [competitor] compare to alternatives" — and uses the answer to form an initial list of three or four names before doing any deeper research. The assistant is doing the job a colleague's recommendation or a quick Google search used to do: narrowing an unbounded category into a short, workable list.
What happens after that first list is still mostly conventional. The buyer visits the named vendors' sites, reads pricing, and often books a demo, exactly as before. The assistant changes who makes the first cut, not who closes the deal.
This also happens earlier in the buying process than most vendors assume. A buyer who has not yet told anyone internally they are shopping will often ask an assistant a rough, half-formed question weeks before a formal evaluation starts, simply because it is faster than a search engine for an open-ended question. That means the shortlist can already be forming before your sales team, or even your marketing analytics, has any record the buyer exists.
What are assistants actually reading to build that list?
Indexed pages, and the third-party pages that discuss a category: review platforms, comparison articles, community discussion and documentation. Google states plainly that its AI features build on the same page eligibility as ordinary search results, with no special extra requirements [1]. The practical implication is that a page invisible to normal search is equally invisible to an AI feature built on top of it.
Each major assistant provider also runs its own crawlers with distinct jobs — training, live retrieval, and the index behind citations — and each is controlled independently in robots.txt [2][3][4]. A site that blocked every bot indiscriminately during an earlier, more defensive era of web security is often invisible to these systems today, and often nobody has checked.
That last point is worth restating because it is the single most common gap we find in a first audit: not bad content, not a technical mess, but a firewall or CDN rule from years ago quietly refusing every crawler with "bot" or "GPT" in its name, set up for a reason nobody now remembers.
| Factor | Why it matters |
|---|---|
| Crawlers allowed in robots.txt | Each assistant provider runs named crawlers with different jobs; blocking one removes you from that specific surface |
| Indexed, reachable pages | Assistants lean on the same eligibility as ordinary search results |
| Third-party mentions | Review sites and comparison posts are what assistants read to name vendors, more than a homepage alone |
| Specific, quotable pages | Pricing, specs and comparisons are easier to lift into an answer than a persuasive marketing page |
Does this change which pages a B2B company needs?
It raises the bar on the pages the rest of this cluster already asks for, rather than adding new ones. A comparison page written to be lifted in two sentences serves an assistant and a human evaluator at once. This matters even more in categories where the product itself does the explaining, which is the case for developer-facing tools — see SEO for developer tools — and for companies where the product's own free tier generates the citable content, covered in product-led SEO.
How do you know whether you are showing up?
Ask the questions your buyers would ask, across several assistants, on a fixed weekly schedule, and record whether you are named and whether you are linked. A single answer varies day to day, so one check tells you little; a tracked rate over weeks tells you whether the gap is closing. Keep the same twenty or so questions every week rather than rewriting them after a bad run, or the trend line stops meaning anything. The full set of numbers worth tracking, alongside this one, is in B2B SEO metrics.
Most B2B teams check whether they rank in Google and never once check what ChatGPT says about their category. That is the cheapest audit available right now, and almost nobody has run it.
Questions people also ask
Do B2B buyers really ask ChatGPT which vendor to use?
Increasingly, yes, especially for an early shortlist. They still verify on the vendor's own site and in a demo, but the first three or four names on their list are now often assistant-generated rather than Google-generated.
What should we allow in robots.txt for AI search?
Allow the major assistants' crawlers explicitly rather than relying on a wildcard, check that your hosting or firewall is not silently blocking them, and confirm with a real fetch that the page still returns full HTML to each one.
Does this replace ordinary SEO for a B2B company?
No. Google says the same foundational SEO practices apply to its own AI features, and the assistants draw heavily on the same indexed, well-structured pages that ordinary SEO already asks for.
Where do assistants get vendor names if not from our site?
Review platforms, comparison articles other people wrote, and community discussion. If those sources do not mention you, an assistant usually will not either, however good your own site is.
Sources
- AI Features and Your Website — Google Search Central (checked 22 September 2026)
- Overview of OpenAI crawlers — OpenAI (checked 22 September 2026)
- Does Anthropic crawl data from the web, and how can site owners block the crawler? — Anthropic (checked 22 September 2026)
- Perplexity crawler bots guide — Perplexity (checked 22 September 2026)
Find out what AI assistants say about your category
Book a free 30-minute consultation with Ibrahem and Jeremy. We look at your site while we talk, and you leave with a prioritised plan.
Who wrote this
Jeremy ZavaCo-founder
Leads growth and client partnerships.
Works on: CRO, SEO
Reviewed before publishing by Ibrahem Helal, Co-founder.
- B2B SEO: the complete guide for 2026A buyer's guide to B2B SEO: how it differs from B2C, which pages produce pipeline, what it costs, how long it takes, and how to measure it in revenue.
- B2B SEO Metrics That Actually MatterThe B2B SEO metrics worth reporting are opportunities, pipeline value and closed revenue by landing page, with rankings and traffic only as leading indicators.
- Product-Led SEO, DefinedProduct-led SEO means treating your app's own surfaces — docs, templates, the free tool, the changelog — as ranking pages, not just your marketing site.
- SEO for Developer ToolsDeveloper tool SEO runs mostly through documentation and API reference pages, not the marketing site. Here's how to make both actually rankable.