White paper · November 2026LIAISON MARKETING

How to rank first in

AI answers

The Shortlist Is Borrowed

How ChatGPT, Gemini and Perplexity choose the firms they name, and ten actionable insights for partners, CEOs and their marketing teams.

Clem Assingo, DirectorValentin Toussaint, Growth DirectorBrazil and Spain

Free. 30 pages, to read here or to download as a PDF. Your name, email and company come to LIAISON and stay with us. Privacy

The paper is open for you.

A woman checking her phone in a dim room
30 pages · PDF and online

Ten insights. Ten ideas for Monday.

Each one says what the research shows, what it means for your firm, and one idea to try on Monday.

  1. 01Named in 69, First in 25

    There is no fixed number one. Ask twice and the list changes, so count how often you are named.

  2. 022.9%

    Each assistant names different firms. Three assistants agreed on 2.9% of the firms they named.

  3. 03Hundreds

    To be named by AI, first be found in search. Google’s AI takes its pages from Google’s own index.

  4. 0415.2%

    AI repeats what others write about you. A brand’s own pages are 15.2% of what AI cites.

  5. 0542%

    AI takes its names from Google and the directories. For law firms, google.com supplied 42% of AI citations.

  6. 064.3 stars

    Good reviews get you on the list, not to the top. Ratings work as a filter.

  7. 0744.2%

    AI quotes the top of the page. 44.2% of ChatGPT’s citations came from the first 30% of a page.

  8. 083 of 54

    Shortcuts to AI answers rarely work. 3 of 54 rewriting tests lifted a page beyond chance.

  9. 093%

    Each language and each city has its own list. Regional companies were named in 3% of the searches they fit.

  10. 101.08%

    AI sends few visits, but many people ask it first. 1.08% of visits, yet 46% of US adults would consult AI before a lawyer.

For partners, CEOs and their marketing teams in professional services

A marketing study, from what a buyer can see.

More buyers now ask ChatGPT, Gemini or Perplexity which firm to hire. We wanted to know how the assistants choose, and what a firm can do to be named. We asked them ourselves, then read the 2025 and 2026 research and what Google, Microsoft, OpenAI and Perplexity say about their own systems. The short answer: an assistant borrows its shortlist. It names the firms that search finds and that others write about. Each of the ten insights says what the research shows, what it means for your firm, and one idea to try on Monday. We see what a buyer can see, not what happens inside your firm, so each idea is a suggestion to weigh against what you know.

The work behind this paper

  • 3assistants asked the same question, 11 November 2026
  • 10firms named, and every firm site and listing they cited opened
  • 12studies from 2025 and 2026, read at the source
  • 3research papers, one accepted at NeurIPS
  • 7pages of documentation by Google, Microsoft, OpenAI and Perplexity
  • 3talks and one lawyers’ forum thread, read in full
  • 2free reports that show AI visibility, by Google and Bing
  • 30sources, each one dated, and linked where public

Written by

The two people who run LIAISON.

  • Clem Assingo

    Clem Assingo

    Director, Foz do Iguaçu

    He has lived in six countries and opened companies on more than one continent. He is the first reader of everything that goes out under a client’s name.

  • Valentin Toussaint

    Valentin Toussaint

    Growth Director, Madrid

    An expatriate who made his life in Spain, he runs search, paid media and measurement: the engines that find the clients who are already looking.

The cover of the white paper: How to rank first in AI answers, The Shortlist Is Borrowed

Free · 30 pages

Read it tonight.

On this site, set for a phone or a screen, or as the PDF.

Free. 30 pages, to read here or to download as a PDF. Your name, email and company come to LIAISON and stay with us. Privacy

The paper is open for you.