How B2B Buyers Choose Vendors Now: the Shortlist Forms Inside the AI Chat

Most B2B buyers now begin supplier research by asking an AI tool which providers are worth considering. The model returns a handful of names, and that answer becomes the shortlist. The whole exchange happens before anyone visits a website. How AI changed B2B buying has little to do with the traffic numbers your team watches each week.

This piece sets out the new buying sequence, the data behind it, and the reason your analytics cannot see the moment that now decides most deals.

How do B2B buyers use AI to choose suppliers?

B2B buyers increasingly start supplier research inside an AI chat rather than a search engine. They describe their requirement in plain language, and the model returns a small set of named providers. That set becomes the working shortlist, refined with follow-up questions in the same session, before any vendor's website records a visit.

This moves the decisive moment earlier, into a conversation no vendor sees. A few years ago, research meant typing a query into Google and clicking the three or four links that looked closest. Every one of those clicks landed in someone's analytics. The vendor could see the visit, attribute it to a source and pick the buyer up further down the funnel.

Now the first move is a question put to a model, and the model answers with names. The buyer reads a synthesised recommendation rather than a list of links. They may never reach a results page. The shortlist exists before any site logs a session, and it is built from whatever the model has learned to associate with the problem the buyer described.

That changes who gets considered. If the model names three providers and yours is not among them, you are out of contention. You were filtered out at a stage you could not observe and did not know was happening.

The new buying sequence, step by step

The journey now runs through the AI tool first and the vendor's website second, if at all. The buyer forms a view of the market before any salesperson, and often before any marketer, has a chance to shape it.

A typical sequence looks like this. The buyer asks the model to recommend providers for a specific use case in their sector. They get back a handful of names with brief reasoning. They ask follow-up questions to narrow the field, probing fit for a regulated environment or integration with a tool they already run. The model refines the list. By the end of a single session, the buyer holds a shortlist and a rough ranking, assembled entirely from the model's synthesis.

When that buyer eventually contacts sales, they arrive with a formed opinion. They are checking what they already believe rather than exploring options. The vendor who was named first, and described most favourably, walked into the conversation with an advantage that no demo can fully reverse.

How AI changed B2B buying: what the 2026 data shows

How AI changed B2B buying comes down to one shift: the supplier shortlist now forms inside the chat. AI tools have become a primary channel for B2B supplier research, and the answers they give now change which vendors buyers consider and ultimately choose. The data below, from named 2025 and 2026 sources, describes current behaviour rather than a forecast.

The headline numbers come from a buyer study G2 published in 2026, surveying more than a thousand B2B buyers. According to G2's 2026 buyer study, a majority now begin research with AI chatbots more often than with a traditional search engine, a sharp rise on the position a year earlier. The same study reports that a large share of buyers chose a different vendor than they had first intended after AI guidance, and that a meaningful proportion bought from a vendor they had never heard of before the AI named it. These are the figures we are still cross-checking against G2's published release, so treat them as directional rather than precise, but the direction is not in doubt.

Other sources point the same way. Responsive, drawing on Digital Commerce 360 data in 2025, found that roughly one in four B2B buyers now use generative AI more than conventional search for supplier research. On the search side, Seer Interactive reported in 2025 that organic click-through to the top result falls by around 60% once an AI answer sits above it, and Pew Research Center found in 2025 that people click through to a website just 8% of the time when an AI summary is shown.

Here is how the picture holds together.

What the data shows	Source and year	What it means for your funnel
Most B2B buyers now start research with AI chatbots more than Google	G2 2026 buyer study (figures pending final verification)	First impressions of your market form inside the chat, not on your site
Many buyers changed their chosen vendor after AI guidance	G2 2026 buyer study (figures pending final verification)	The AI's recommendation overrides a buyer's prior intent
Around 1 in 4 B2B buyers use generative AI more than conventional search for supplier research	Responsive / Digital Commerce 360, 2025	A quarter of your prospects are sourcing names from a channel you do not track
Organic click-through to the top result falls ~60% when an AI answer sits above it	Seer Interactive, 2025	Ranking first earns far fewer visits than it used to
People click through to a site just 8% of the time when an AI summary shows	Pew Research Center, 2025	The buyer now reads the answer in place of the page, so the decision starts there

Taken together, these figures describe a buying process where the influential moment sits inside an AI conversation your team never sees.

Why your analytics cannot see the shortlist forming

Your analytics record visits to your website. The shortlist is built before any visit, so the moment that decides whether you are considered leaves no trace in your reporting. You see the prospects who arrive already aware of you, and none of the ones the model never named.

This is the gap behind the objection that search is not a big part of your funnel. It may well be true that organic search drives a small slice of your tracked pipeline. That tells you nothing about the conversations happening in AI tools, because those conversations do not produce a tracked session. A buyer can ask a model about your category every day for a month, build a shortlist that excludes you, and contact one of your competitors, while your dashboards show none of it. The funnel you measure is the part that survived a selection step you had no visibility into.

The same question gets put to AI tools many times a day across any active B2B category. Each one hands the introduction to whoever the model decided to name. By the time a prospect reaches your sales team, the decision is half made, and you had no part in shaping it. The deals you can see are the ones that got past the filter. The ones that never reached you never appear at all.

A channel you cannot measure is still selecting your prospects for you, and the absent deals are the real cost.

Where Forge fits

Forge runs generative engine optimisation as a programme for regulated B2B companies, with the aim of moving you from absent to recommended in the AI answers your buyers actually read. The work starts with an audit. We ask AI the questions your buyers are asking and record exactly where you stand, noting which questions already name you and which sit open with no clear leader. From there we build the content the models draw on and earn the citations and structured signals that teach them to trust your name, then re-score the movement month by month.

We have seen what that movement looks like. A UK digital identity platform we work with grew its monthly AI reach from around 39,000 to 2.9 million over six months and won a seven-figure contract directly through AI search. A healthcare compliance platform saw its largest contract arrive through ChatGPT, with AI-sourced pipeline rising over the same period. Across the live programmes we run, the pattern is consistent: categories that started with no named leader, moving toward a clear one.

If you want to understand the mechanism behind this in more depth, our explainer on what generative engine optimisation is covers the discipline end to end, and our guide to measuring how often AI recommends your brand sets out the metric this all turns on.

Frequently asked questions

How do B2B buyers use AI to choose suppliers?

Buyers describe their requirement to an AI tool and receive a short set of named providers, which becomes their working shortlist. They refine it with follow-up questions about sector fit, scale and integrations, often within one session. By the time they contact a vendor, they hold a formed view assembled from the model's synthesis rather than from any single supplier's site.

How has AI changed the B2B buying journey?

AI has moved the first and most influential step of research out of search engines and into chat. The vendor shortlist now forms before a buyer visits any website, built from what the model associates with the buyer's problem. Vendors not named at that stage are filtered out before they know a deal exists.

Why can't our analytics see this?

The selection happens inside an AI conversation that produces no visit to your site, so it leaves no record in your analytics. You see prospects who arrive already aware of you, but not the larger group the model never named. The step that decides consideration happens entirely outside anything your reporting can measure.

Search isn't a big part of our funnel, so does this matter?

It matters precisely because the effect is invisible. Low tracked search traffic tells you nothing about how many buyers research your category in AI tools, since those conversations never become a tracked session. The selection happens in-chat regardless of your search numbers, and it is already shaping deals you never see reach your pipeline.

When does AI influence the decision in a B2B sale?

It influences the earliest stage, when the buyer is deciding who to consider at all. According to G2's 2026 buyer study, many buyers changed their chosen vendor after AI guidance, which means the influence carries through to the final decision and not only the longlist. By the time sales is involved, the AI-shaped shortlist has usually set the terms.

See where you stand today

Book a discovery call and we will run your category live. 30 minutes. No pitch deck. We will ask AI the questions your buyers are asking and show you exactly what it says about you today, including which questions already name you and which sit open.

Start with a free AI-visibility audit at forgetogether.agency/GEO.