November 18, 2026Last updated: November 18, 20268 min read9 sections

Your Buyers' Shortlist Is Written Before They Visit Your Site

By Tom Hunt, Founder & CEO, Fame

Tom founded Fame in 2019 and leads the company today, including Fame Search, its AI search agency for B2B companies. LinkedIn

Key Takeaways

  • Buyers who ask an AI assistant first arrive with a shortlist already in hand.
  • Ranking still matters, but it appears to buy a place in the pool of sources and not the click.
  • The answer decides who is listed, how each company is framed and which facts are stated.
  • Referral traffic understates this, so track share of answers and ask buyers how they heard about you.
  • Your site now serves two readers: the engine before the shortlist and the buyer confirming it afterward.

More B2B buyers now appear to ask an AI assistant who can solve their problem before they open a single vendor website. The answer names a handful of companies, and those companies are the shortlist. Ranking first on Google is still useful, but it is no longer the finish line, which is the whole reason the AEO agency category exists.

I made the first version of this argument in the question is no longer "do we rank?". This piece takes it one step further: if the shortlist is written somewhere else, what is your website for, and what should a marketing team do differently?

What changed about how buyers build a shortlist?

The old sequence was search, click, compare. A buyer typed a few words into Google, opened five tabs and built the shortlist themselves from what they read. Your job was to be one of the tabs.

The new sequence puts a summary in the middle. The buyer describes their situation to an assistant, often at length. HubSpot says prompts now run 40 to 100 words, with buyers describing their whole situation. The assistant does the reading and comes back with names. The buyer may then visit two or three of those sites to confirm what they were told.

None of this needs the buyer to trust the assistant completely. It only needs them to use it for the first pass, the way they once used the first page of results. A shortlist is a filter, and whoever does the filtering shapes the outcome.

The click data points the same way, although none of it is specific to B2B:

  • Ahrefs, analyzing 300,000 keywords: the top-ranking page loses 58% of its clicks when an AI Overview appears (December 2025 data; 34.5% in the April 2025 run).
  • Pew Research: users click a standard result about half as often when an AI summary appears (8% vs 15%), and only 1% click links inside the summary.
  • HubSpot forecasts 95% of buyer journeys will start with a large language model within two years. That is a forecast from a company with a view on the subject, so hold it loosely.

You do not have to believe the largest of those numbers to accept the direction. Some share of your buyers now arrive with a list already in hand, and that share appears to be growing.

Is ranking first still worth it?

Yes. I am not arguing that SEO is dead, and I would be suspicious of anyone who does. On the best available data, ranking and citation are linked. Ahrefs' analysis reported that 76% of 1.9 million AI Overview citations also rank in the traditional top 10. Its stability study found 11% of ChatGPT citations lasted seven or more days when the page ranked in Google, against 4% when it did not. These are correlations from a company that sells SEO tools, but they point the same way.

What has changed is what ranking buys you. It used to buy the click. Now it appears to buy a place in the pool of pages an engine might draw on, and after that a second selection happens that you cannot see in a rank tracker. ChatGPT in particular does not simply copy Google: in an Ahrefs study from September 2025, about 10% of ChatGPT's cited URLs on short-tail queries overlapped with Google's top 10.

So ranking is the qualifying round. Treating it as the final is how a team ends up with a strong rank report and a competitor's name in the answer.

What does the answer decide before a buyer reaches you?

Three things, and each used to be decided on your site.

Who is on the list. If you are not named, you are not compared. The buyer does not know what they did not see.

How you are framed. The answer often attaches a phrase to each name: best for small teams, the expensive option, strong on compliance. That phrase may come from your own page, from a list someone else wrote or from a review. You might agree with it or you might not.

What the facts are. Price, who you serve, what you integrate with. If an engine states your price wrongly, the buyer may rule you out without ever checking. This is why a tracking panel should include fact-check questions alongside discovery ones, and why the first fix for an error is your own page, followed by the third-party page that likely caused it.

There is a further wrinkle. Being cited and being named are different things. Ahrefs found that in 54% of cases on ChatGPT a site was cited without the brand being named. Your page can do the work of informing the answer while a competitor gets the recommendation.

Why does your analytics miss this?

Because the influence happens before the visit, and often instead of it. Many buyers read an AI answer without clicking anything.

Our own site is a fair example. At its October 30, 2026 snapshot, fame.so logged 41 AI assistant visits in 28 days against 42,966 Google impressions. If I judged AI search by referral traffic alone, I would conclude it barely exists. Yet the share of tracked AI answers citing fame.so went from 38% on June 1 to 76% on October 30, 2026. Both numbers are true. One counts clicks and the other counts presence in the answer.

That gap is why I would add a required free-text field to your demo form asking how the buyer heard about you. Self-reported data has its own flaws, but it catches some of what analytics miss.

What is your website for now?

It has two readers where it used to have one.

ReaderWhen they arriveWhat they need from the page
The engineBefore the shortlist, while it is assembling an answerPlain statements it can lift: what you do, who it is for, what it costs, how you compare
The buyerAfter the shortlist, to confirm what they were toldThe same facts, matching the answer, plus proof and a clear next step

The practical consequence is that vague pages now cost you twice. A homepage that says you "empower teams to do more" gives an engine nothing to quote, and it gives a buyer who arrives from an answer nothing that confirms it.

The pages that serve both readers are unglamorous: a category page that says what you do and who it is for in the first lines, a pricing page with real numbers, comparison pages that are fair about where the other option is better, and answer pages for the specific questions buyers ask before talking to sales. We cover the writing method in how to write an answer page that AI assistants quote.

Where is the shortlist actually written?

Partly on your site and partly off it, and the published numbers disagree about the split. OtterlyAI's index put company-owned pages at 53.7% of citations across engines. Other vendor studies put far more weight on third-party sources. They measure different things, so the safe plan is to work both sides.

On your side, that means the pages above. Off your site, it means the places an engine appears to look when asked for options: "best of" lists, review sites and directories, podcasts and videos where someone else introduces you. A mention on someone else's site, in someone else's words, is a different kind of evidence from anything you publish yourself.

It also differs by engine. The brand that ChatGPT names is not always the one Perplexity or Gemini names, which is why we report each one separately. See how ChatGPT, Perplexity and Gemini differ.

What should you do this quarter?

  1. Write down 20 questions a buyer asks before they talk to sales, in their words. Keep most of them unbranded.
  2. Ask each one several times in ChatGPT, Perplexity, Claude and Gemini. Record who is named, who is cited and what is said about you. Our tracking method has the spreadsheet columns.
  3. Check that the AI search crawlers can reach your site.
  4. For each question where you are missing, open the pages that were cited instead. That list is your work plan.
  5. Fix your own pages first: direct answers, real prices, fair comparisons.
  6. Then go where the answers point off your site, whether that is a list, a review profile or a show.
  7. Repeat the tracking every week and read trends over several weeks, not single answers.

None of this is difficult. It stalls when nobody owns it, so give the weekly check to one named person and put thirty minutes for it in their calendar. If you want the long version, read the complete guide to B2B AI search.

What do we not know yet?

Quite a lot, and I would rather say so than pretend otherwise.

  • How large the AI-first share of B2B buying really is. The forecasts are forecasts, and most click studies cover consumer or general search.
  • How each engine chooses sources. None of them publishes a formula, and a model update can shift the pattern without any change on your side.
  • How well any of this transfers to your category. The only test that counts is your own questions, run repeatedly.
  • What it does for clients of ours. Fame has no client results to report yet. The only result we have is our own site, and our numbers do not predict yours.

What I am confident about is narrower. For the buyers who do ask an assistant first, the comparison you used to win on your website now happens before they arrive. It is worth knowing whether you are in it.

Want your company named in AI answers? Fame Search is our AI search agency for B2B companies: weekly tracking across four AI engines and one action on your site every day, for $2,000 a month.

Share thisLinkedInEmail

Frequently Asked Questions

  • No. On the best available data, pages that rank in Google are more likely to be cited and to stay cited. Ranking is now the qualifying round, and a second selection happens inside the AI answer.

Want us to run this for you?

Fame starts and grows B2B podcasts that generate pipeline. Tell us who you are trying to reach and we will send a costed proposal within 1 working day.