Are buyers shortlisting you in AI before they call?

Yes, and more than most leadership teams realise. B2B buyers have long done most of their deciding before they pick up the phone. What has changed in the last two years is where that deciding happens. A buyer who once typed three words into Google, clicked five websites and rang two of them now types a full paragraph into ChatGPT or Google’s AI Mode and gets a written answer with a handful of names in it. If yours is not among them, the buyer moves on and your analytics record nothing.

I made this point to a room of West Yorkshire business owners at the Digital Knowledge Exchange conference in Leeds on 14 October. Here is the fuller version, with the numbers checked.

How early do buyers decide?

Buyers are well over halfway through their decision before they speak to a supplier, and they rarely buy from anyone who was not on their list at the start.

The same research found buyers were about 61% of the way through their journey at first contact, that they started 79% of those conversations themselves, and that the day-one shortlist held 3.6 suppliers. Read those together and the picture is uncomfortable. By the time someone gets in touch, the list is short, it was set early, and you did not get a vote.

Referrals have always worked this way. What is new is that the early research, the bit where the list gets formed, has moved somewhere you cannot watch.

What has AI changed about the shortlist?

AI has turned the early research from a list of links into a written recommendation. The buyer no longer compares ten websites themselves. They describe their problem, and the tool does the comparing and hands back a shortlist with reasons attached.

You can see it in how people search. Semrush found that ChatGPT prompts sent without its search feature averaged 23 words, across 80 million clickstream records from late 2024. Twenty-three words is a brief: the trade, the town, the size of business, the budget, the worry. The AI matches that brief to the businesses whose websites, reviews and articles describe them clearly enough to be matched.

Where does the shortlist get made now?

In three places, and they are one job rather than three. Google’s normal results still matter, AI summaries now sit on top of a large share of searches, and chat tools answer the longer, more considered questions.

Every figure checked against its primary source, October 2026. Vendor studies, so read them as direction rather than precision. On a phone, scroll the table sideways.

The last column is the one I would put in front of a board. None of these three is a separate marketing channel with its own budget line. They all draw on the same evidence: a clear website, consistent details, real reviews that name the service, and articles and mentions that show you know your stuff.

Is fewer enquiries a problem?

Not necessarily, and this is the part founders tend to miss. If the shortlisting happens before the call, the people who do call are further along, better informed and arrive with some trust already in place. They have, in effect, been pre-sold by a source they chose.

When they do speak to you, they’re probably far warmer, far further into the funnel, and come with some trust as well. But the question is, are you on the shortlist?

From my closing keynote, Digital Knowledge Exchange conference, Leeds, October 2026

The early data points the same way. Ahrefs reported that visitors arriving from AI search converted at 23 times the rate of ordinary search visitors on its own website, and Semrush estimates an AI search visitor is worth 4.4 times an organic one. Ahrefs’ figure comes from one website, its own, and Ahrefs said itself it was not sure the result would scale; Semrush does not publish how it reached its number. I would not build a forecast on either. I would take the direction seriously.

The risk sits on the other side. If enquiries are falling and nobody can say whether that is fewer buyers or fewer shortlists with your name on them, the business will reach for the wrong fix, usually more spend at the top of the funnel.

What should a leadership team measure instead?

Measure whether you are named, not just whether you are visited. Website sessions still matter, but on their own they will increasingly tell you the wrong story. I covered the shift from clicks to mentions in more depth in an earlier piece on why citations are the real metric. In practice, three things change.

None of this needs a new tool. It needs someone to own the question.

How do you find out if you are on the list?

Ask the machine, then cross-examine it. Type the question your best customer would ask, with your trade and town in it, read the answer, and then ask it to justify the list, including why you are not on it.

When I asked the room in Leeds where they last looked for a supplier, nearly everyone said Google, and roughly half had asked AI as well. One business owner then asked whether their business coming up top was genuine or just personalised to their phone. That is the right instinct. What a buyer sees is not what you see from your own logged-in, much-visited browser, so check from the buyer’s side.

I have seen the same thing turn a conversation in a client relationship. A long-standing B2B technology client asked, almost in passing, how far we had got with optimising for AI. A live prompt comparing how the tools described them against their competitors moved the project on the spot, from rebuilding a website to rebuilding it with AI visibility designed into every page. In the 90 days after the new site launched, sessions were down 25.5% on the previous 90 days while engagement rate was up 55%. Fewer visits, more of them from people who meant it. That is the shortlist pattern in miniature.

If you want the wider version of that audit, I set out four lenses in what AI is already telling buyers about your business.

FAQ

Do B2B buyers really use ChatGPT to find suppliers?
A growing number do, alongside Google rather than instead of it. The more important point is that the shortlist was already being formed before first contact. AI tools have made that early research faster, longer and harder for suppliers to see.

If AI answers mean fewer clicks, is SEO still worth paying for?
Yes. AI summaries draw heavily on pages that already rank well, so SEO is the foundation the AI work sits on. What changes is how you judge it: by whether you are named on the questions that matter, not by sessions alone.

How often should we check whether AI names us?
Every six to eight weeks is enough for most businesses. AI answers vary from run to run, so run each question two or three times and look for the pattern rather than reacting to one result.

Who in the business should own this?
Whoever owns the commercial pipeline, not only whoever owns the website. It is a question about who gets the conversation, which makes it a leadership question.

I am Jonny Ross, a Fractional CMO working with scaling SMEs across the North of England. I have spent 27 years in search and visibility, starting with my own online sunglasses business in 1999, and I now sit in leadership teams helping them turn visibility into pipeline.