Your buyer's shortlist is already built

By the time you get the call, the list is made. Increasingly, an AI assistant helped make it.

Every part of the way we market and sell assumes a customer who is looking for you. Search, advertising, events, direct outreach and content all rest on the same premise: the customer starts by looking for options, and the seller's job is to be found, or to reach the customer first. That premise has expired.

Buyers now put a different question to an assistant before they speak to anyone: who should I consider? The assistant summarises, compares and recommends. It writes the shortlist before any of your activity has a chance to run. Everything after that point, including the call, the proposal and the pitch, works on a list that already exists. Good selling improves your odds against the names on the list. It can't put you on it.

The sale doesn't start at the first call. It starts when the list is written.

The profession's answer is precedent

Most of the response so far has been a translation. SEO becomes AEO or GEO. The website stays a website. The buying process stays the one we already knew, with human involvement arriving a little later. A new problem, solved with the familiar answer.

The more useful question is what a web presence is for when its first reader is an assistant. Websites were built as destinations for people who arrived by clicking. For a growing share of buyers, the first reader summarises, compares and recommends before anyone clicks. That is a different job, and it is rarely put to the people who built the destination.

This is not a criticism of web teams, in-house or agency. The brief changed under them. But asking a team skilled in building destinations to design for a reader that may never visit is a bit like asking a bricklayer to start building in glass. The materials, the skills and the mindset are different. The capability has to be added, not just the label changed.

There is a second risk, and it is structural. The people best placed to help look like the ones who built the last system. Web, search and SEO is an industry with its own tools, job titles, retainers and reporting. An industry like that has every incentive to rebuild itself under a new name: the tools gain a letter, the retainer gains a deliverable, and the buying process stays the one it already knew. Nobody needs bad intent for this to happen, because incentives and experience do it on their own. It is hard to reimagine the buying experience when you were hired to optimise the one that exists.

We are not exempt. We sell work in this space too. The test, for us and for anyone you hire, is whether the work changes what the buyer experiences or only what the report is called.

What the evidence shows

We are not the only ones counting. The numbers below come from different markets, so each one is labelled.

The pattern predates AI. In 6sense's 2025 B2B Buyer Experience Report (just under 4,000 buyers), the vendor that won the deal was already on the buyer's shortlist on Day One of formal evaluation 95% of the time. Buying groups had filled four of five shortlist spots on Day One.

What has changed is who writes the list. G2's Answer Economy research (a March 2026 survey of 1,076 B2B software buyers and decision-makers, published April 2026) found 51% now start their research in an AI chatbot more often than in Google. Eleven months earlier it was 29%. G2 reports AI chatbots as the number one source influencing which vendors make the shortlist. 69% of buyers said chatbot guidance led them to choose a different vendor from the one they had planned on, and 33% bought from a vendor they were not familiar with.

Consumer behaviour points the same way. Similarweb tracked US desktop users from July to December 2025 who asked ChatGPT a question in finance, travel or beauty and received a specific brand recommendation. Within seven days they were 2.5 times more likely to visit the recommended brand than a direct competitor. Users who had visited that brand in the previous four weeks, or named it in their prompt, were excluded.

The UK picture is more measured. Ofcom's November 2025 discussion paper reports that 32% of UK online adults visited at least one AI chatbot in June 2025, with ChatGPT reaching 26% (Ipsos iris, which measures visits, not search use). Ofcom's own small interview study of 24 UK adults, run by YouGov, found people still mostly start in Google, use AI for low-stakes and exploratory questions, and check the answer before acting on anything that matters. The authors call the findings exploratory. Both facts hold at once. The list is increasingly written by assistants, and people still verify it. Being on the list is necessary, not sufficient.

Pew Research Center's US browsing data, as cited by Ofcom, shows what that does to the web: when Google showed an AI summary, users clicked a link on 8% of visits, against 15% when it did not.

Every figure here comes from a specific market. 6sense and G2 measure B2B buying, and G2 only software buyers. Similarweb measures three consumer categories, on one assistant, in one country. Nobody has measured every sector the same way, so read each number as direction, not as a precise rate for your market. The mechanism does not depend on the sector, because the reader summarises, compares and recommends before anyone clicks.

The list is also unstable. SparkToro and Gumshoe asked 12 identical prompts across ChatGPT, Claude and Google's AI nearly 3,000 times, using 600 volunteers. The odds of getting the same list of brands twice were under 1 in 100. A shortlist from an assistant is a probability, not a ranking. You can't check it once and relax. You have to measure it repeatedly, the way you would the weather.

Winning the list

The contact points that matter in the end are still human: the meeting, the event, the store, the conversation. None of that goes away. But a buyer only steps into them after the list is built. To be front of mind when they do, you have to be on the list the assistants write.

Three things follow from that.

• Find out where you stand. Ask the questions your buyers ask, across several assistants, repeatedly. Record whether you appear, not just where.
• Treat your web presence as reference material for a machine reader as well as a human one. Make specific, consistent, attributable claims, and keep the facts the same wherever they appear, because people check what the assistant said before they act.
• Put the question to people who can answer it. It spans web, data, content and reputation. Ask any supplier, ours included, whether the proposal starts from how buyers now behave or from the service line they already sell.

If you're not on the list, you're not in the running. No amount of effort afterwards fixes that.