Why your brand doesn't show up in ChatGPT
A diagnostic checklist — from blocked crawlers and client-rendered pages to weak category association — in the order worth checking them.
If ChatGPT never mentions your brand, the cause is usually one of a handful of specific, checkable things. This article works through them in diagnostic order, from the fastest to check to the slowest to fix.
First, establish which problem you have
Before checking anything, run two tests. Ask ChatGPT your category question with search enabled, and again with it disabled. The result tells you where to look:
- Absent in both — you have a reputation problem and probably a retrieval problem. Work through everything below.
- Absent with search on, present with it off — a retrieval problem. The model knows you but cannot find or use your pages. Start at check one.
- Present with search on, absent with it off — a reputation problem. Your pages work; the model has no strong association. Skip to check six.
This five-minute test saves weeks of work on the wrong half.
Check one: is your content actually fetchable
Fetch your key pages as raw HTML with a non-browser user agent. If the substantive content only appears after JavaScript runs, retrieval will frequently get an empty page. This is the most common single cause and the most fixable.
Check two: are you blocking the crawlers
Read your robots.txt carefully. OpenAI operates separate crawlers for different purposes, and blocking the search crawler removes you from retrieval entirely. Many sites blocked all AI agents during an earlier wave of concern about training data and never revisited the decision. Confirm any exclusion is still deliberate.
Check three: is bot protection interfering
CDN bot management, rate limiting, and WAF rules frequently challenge or block unfamiliar user agents. From the crawler's perspective this is indistinguishable from the page not existing. Check your edge logs for blocked or challenged requests from AI crawler user agents.
Check four: are you indexed and relevant
Retrieval runs through a search index. If your page is not indexed, or is not relevant for the keyword-shaped rewrite of the user's question, it is never a candidate. Check index coverage and search for the literal query form.
Check five: is there anything extractable on the page
If the page is fetched but the answer is buried in paragraph nine, or every section opens with context before resolving, there may be nothing for the model to lift. Read only the first two sentences under each heading and see whether they answer anything. That approximates what extraction sees.
Check six: does the wider web describe you clearly
If the model has no strong association with your brand, the cause is off-site. Search for your category and read what the top third-party sources actually say. Common findings: you are absent from every roundup, you are described using outdated positioning, or you are described five different ways across sources so no confident category placement is possible.
Check seven: are you asking the right question
Sometimes the brand is fine and the test is wrong. A question phrased in language nobody uses, or one so broad that only three global brands could plausibly be named, will not surface a mid-market vendor no matter what you do. Check that your test questions match how buyers actually ask.
Check eight: is it just variance
Generated answers differ between runs. A single absence proves nothing. Run each question five to ten times before concluding you are missing, and measure the rate rather than the instance.
What to fix first
- Anything in checks one to three, because they invalidate everything downstream and are usually a same-week fix.
- Index coverage and page structure, which typically show movement within two to six weeks.
- The off-site record, which is the slowest but is what separates occasional citation from consistent recommendation.
Common questions when your brand is missing
- My competitor appears and we are bigger. Why? Market share is not an input. Models learn from how consistently the web describes brands, so a smaller competitor written about clearly and often in your category can outrank you in answers.
- We rank first in Google but never appear in ChatGPT. What is wrong? Almost always one of two things: the page is not fetchable by the search crawler, or the answer is buried rather than stated under a heading. Check fetchability first, since it is binary and quick to test.
- We used to appear and now we do not. What changed? Check three things in order — a robots.txt or CDN bot rule that changed under you, a site migration that altered rendering, and a competitor publishing something better on the same question. Model updates can also shift answers overnight, so check release timing before assuming it was you.
- Does a low domain authority explain it? It contributes, because engines favour established sources, but it is rarely the whole story and it is not something you can fix this quarter. Work through the mechanical checks first — they are fixable now.
- How many times should I test before concluding we are missing? Five to ten runs of the same question. Single absences are within normal variance.
Tip: before starting any of this, ask ChatGPT to read one of your URLs and summarise it. If the summary is vague, wrong, or it reports being unable to access the page, you have found your problem in under a minute.
Still stuck? We typically reply within 1 business day.
Contact support