Guide

Why AI recommends your competitor and not you

Updated November 2026

When a buyer asks an AI assistant what to buy and it names your competitor, the cause is rarely that the model prefers them. With web search on, Claude, ChatGPT, Gemini, and Grok each run a few searches, read the top results, and summarize them. If your competitor is on those pages and you are not, they are in the answer and you are not. Below is what that looks like in real answers, and what you can change.

What the answers are built from

Saidly runs a public weekly watch on one buyer question: "best project management software for small teams". Here is what came back on October 26 and November 2, 2026.

Top three per model, two weekly runs
ModelOctober 26November 2
ChatGPTAsana, Trello, ClickUpAsana, Trello, ClickUp
GeminiAsana, Trello, ClickUpAsana, Trello, ClickUp
GrokClickUp, Asana, monday.comClickUp, Asana, Trello
ClaudePlaky, ONES, Wrikemonday.com, Asana, ClickUp

Asana, Trello, and ClickUp appeared somewhere in every model's answer both weeks. The sources the models cited explain why. They were almost all "best project management software for small teams" roundups (project-management.com, Capterra, The Digital Project Manager, and blogs run by other software vendors) plus the pricing pages of Asana and ClickUp. The three brands at the top are the three that show up on nearly every one of those lists.

Claude's October 26 answer shows the other side. It opened with Plaky, ONES, Wrike, and Rock, and named them as a caution: "several of these lists come from companies that sell project tools (Plaky, ONES, Wrike, Rock) and often rank their own product first." Publishing your own roundup put those vendors in the answer, but not in a flattering way. A week later, with different search results, Claude's list looked like everyone else's.

Why a model names your competitor

You are not on the pages it reads
This is the most common cause. If the roundups, review sites, and comparison articles that rank for your buyer's question do not mention you, a model that summarizes them will not either. Your own homepage is one page among many, and the model knows you wrote it.
Your site does not say what you are in plain words
A model matches the buyer's words to pages that use the same words. If your homepage says "reimagine collaboration" and never says "project management software for small teams" and a price, it is hard to quote. How to get cited in AI answers covers what makes a page quotable.
The model is not sure which company you are
Names collide. In Saidly's own report on November 1, 2026, three models described Saidly correctly and Claude described two unrelated projects with the same name. When a model cannot tell which company a page is about, it may leave you out or mix you up with someone else. Consistent naming, an About page that says who you are and are not, and listings on outside sites all help.
Your site is cited but your brand is not named
Sometimes the model uses your page as a source and still recommends someone else. We wrote about that in why AI cites your site but never names your brand.
You sell a service, and the first answer does not name anyone
Service questions behave differently. In Saidly's October 2026 testing, "help with quarterly estimated taxes for my small business" got a how-to answer from all four models and no firm names. When the same conversation followed up by asking for a firm in Columbus, Ohio, all four named local CPA firms. If you sell a service, the follow-up question is where you get recommended.

What to do about it

  1. Ask the question your buyers ask, on all four models. Write down who is named, in what order, and which pages are cited. The free check does this for one model with no account.
  2. Get onto the cited pages. Ask for inclusion in the roundups that keep appearing, and set up profiles on the review sites the models cite for your category. This is slow. In the data above, being on those lists is what the top three brands have in common.
  3. Make your own pages easy to quote. Say what you sell, who it is for, and what it costs in plain words near the top of the page.
  4. Watch it over time. Claude's list changed almost completely in one week. One snapshot can mislead you in either direction.

One caution: do not try to fake your way onto the list. A marketer got ChatGPT to recommend a deodorant brand that does not exist, which shows how thin the evidence behind these answers can be. It is also not something a real brand can build on. The story is in fake brands in AI answers.

Common questions

Why does ChatGPT recommend my competitor instead of me?

Usually because your competitor appears on the pages ChatGPT reads when it searches the buyer's question, and you do not. With web search on, the answer is built from a handful of results, mostly 'best of' roundups, review sites, and pricing pages. In two weeks of Saidly's project management software watch, ChatGPT gave the same top three both weeks, and its cited sources were roundup articles and vendor pricing pages.

How do I get ChatGPT to recommend my business?

Find out which pages the models cite for your buyer's question, then get onto those pages: review sites, industry roundups, and comparison articles. Make your own site say plainly what you sell, who it is for, and what it costs. For a service business, also check the follow-up question, because a model asked how to do something often explains it without naming anyone, and only names firms when asked who could help.

Do AI recommendations change from week to week?

Yes, and some models change more than others. In Saidly's project management watch, ChatGPT's top three stayed the same across two weekly runs, while Claude's list changed almost completely between October 26 and November 2, 2026. One check is a snapshot, so it is worth tracking the same question over time.

Can I see which brands AI names for my category for free?

Yes. Saidly's free check at saidly.ai/check has a buyer-question mode: type the question your customers ask and see the ranked brands one model names. It needs no account. A paid category watch runs the same question on Claude, ChatGPT, Gemini, and Grok every week.

See your own category

Type the question your customers ask into the free check and see which brands one model names. To track all four models every week, with alerts when a position moves, start a trial and add a category watch.

No credit card to start. Saidly is built by WoodFire Digital LLC.