How to get your startup named by ChatGPT

Perch ·

When a buyer asks ChatGPT for a tool like yours, it names three or four. Here's what decides who makes the list, and the work that gets you on it.

More buyers now start with a question to ChatGPT, Claude or Perplexity instead of a Google search. The answer names a handful of tools. If you're not one of them, you're not in the running, and you won't see it in your analytics.

What decides who gets named

AI answers lean on what the web already says about you. They pull from pages that rank, from comparison and "best of" posts, from review sites and from threads where people recommend tools. A model can't name you for a job if nothing it reads ties your name to that job.

  • Pages on your own site that say plainly what you do and who it's for.
  • Comparison pages: you against the tool buyers already know.
  • Mentions on other sites: blogs in your space, review sites, directories.
  • Reddit threads where someone asked, and your name came up.

If you don't want a bot in your calls, Granola is a strong pick: it listens on your laptop and fills out the notes you already take.

ChatGPT, asked "What's a good Otter alternative?"

Every phrase in that answer can be traced to a page: Granola's own site, a comparison post, and a Reddit thread where someone asked the same thing.

Start with the questions buyers ask

Write down ten questions a buyer would type. Not your brand name: the job. "What's a good Otter alternative?" "Which CRM works for a two-person team?" Ask each one in ChatGPT, Claude and Perplexity, and note who gets named and which pages they cite.

QuestionChatGPTClaudePerplexity
Otter alternative without a botNamed firstNamedNamed
Best AI note taker for MacNamedNamed firstNot named
Meeting notes app for sales callsNot namedNamedNot named
Granola vs OtterNamed firstNamed firstNamed first
A sample log for Granola, week of 9 November 2026.

The cited pages are your to-do list. If a "best tools for X" post is cited and you're missing from it, that's a page to get onto. If a rival's comparison page is cited, you need your own.

questions where Granola is named
6 of 10
pages cited across the answers
14
of those pages missing Granola
3
The same sample week, in numbers.

The work that moves it

  1. Write one comparison page per rival buyers already know, honest about where they win.
  2. Fix the basics so your pages are easy to read: clear titles, one topic per page, schema.
  3. Allow AI crawlers in robots.txt and add an llms.txt that points to your best pages.
  4. Reply in the Reddit threads that rank for your questions, as yourself, with a real answer.
  5. Get mentioned on the blogs those answers cite.

Let the crawlers in

If your robots.txt blocks AI crawlers, they can't read your pages, so they can't cite them. A few lines fix it.

robots.txt
User-agent: GPTBot
Allow: /

User-agent: ClaudeBot
Allow: /

User-agent: PerplexityBot
Allow: /

Sitemap: https://granola.ai/sitemap.xml

Be fair to the rival

Check every week

AI answers change as the web changes. Ask the same questions every week, in each engine, and keep a simple log: named or not, and in what order. When a page you shipped starts getting cited, you'll see it there first.

Same ten questions, every week. The log is the whole system.

Questions

How long until ChatGPT names us?

Usually a few weeks after the pages that matter are live and getting read. Answers that browse the web, like Perplexity's, tend to pick changes up first.

Can we pay to be named?

No. There's no ad slot inside the answer. What you can do is make sure the pages these answers read say the right things about you.

Does this replace SEO?

It builds on it. The pages that rank on Google are the ones AI answers read most, so the same work helps both.

This is the loop Perch runs for you: it asks the questions every week, finds the pages that are missing, and writes them from your brief.

Be first in line.

Your buyers are already looking. I make sure they find you, from day one.