Why does ChatGPT not mention us?
Broadcastwell measures and fixes whether AI search engines name B2B software companies.
Run the free checkStart with the dated answer and its context to investigate an omission. Our records retain named products and explicit source links; they do not expose every source an engine considered or establish why it omitted a company. The September Index keeps ChatGPT, Claude, Perplexity, Google AI Overviews and Google AI Mode separate, so an observation from one surface should not stand in for them all.
The four checks below organize possible explanations. Our answer records do not establish how common each explanation is or whether a particular change will affect the next answer. More than one check may be relevant to the same page.
1. Check the visible sources for your company name
Open the sources explicitly cited in the retained answer, whether they are comparison articles, category roundups, review sites or other publications. Check whether they name the intended company and support the statements at issue. The visible list does not reveal every source considered, and a missing name on those pages does not explain the omission by itself.
We can check a specific public statement that is inaccurate or leaves out a fact, and the evidence needed to correct it. A later naming or citation result needs its own dated observation; a source change alone does not establish that result.
2. Compare your public company facts
Compare the company and product names, category and official website across the public records concerned. Conflicting statements can be documented and checked against dated evidence. An omitted name does not reveal an engine’s internal confidence or prove that identity confusion caused the omission.
We can correct supported company facts on pages we control, following the relevant provider’s documented requirements. Keep the source and date with each correction. We have no measured confidence increase or assured naming result from making those statements consistent.
If you are preparing for OpenAI's reported ChatGPT Business Hub, our guide to claiming your business in ChatGPT lists the name, structured data and facts-file checks to make first.
3. Check access for the relevant page and requester
A failed request and a page opening in a human browser are different observations. Check the intended URL, the requesting service’s documented identity, the response and any applicable access rule. A user-agent label alone does not authenticate the service. Our answer records do not establish how often access failures cause omissions or how often other tools miss them.
We can check the response to a specific verified request against the intended access policy. Where that evidence supports an authorized rule change, read the result back and report access separately from later naming or citation outcomes.
4. Compare the question types in your retained answers
Count naming and citation under the stated rule for each retained question. An explanatory question and a supplier-shortlist question can return different products and sources. Compare their actual answers before describing that pattern. Those records do not establish what happened to a buyer’s pipeline or sales.
We can check which questions produced answers naming or omitting your company, with the engine, valid answer base and collection date recorded.
How to record what you find
Start with fixed buyer questions and repeated, dated answers, retaining the named vendors and explicit source links. Those records describe the outputs to inspect. Access checks and source-fact checks need their own evidence; repeated answers alone do not identify a cause. That is what the Absence Index records for its 14 categories, and what the Fix Sprint ($3,900) measures in its first phase for your own category: ten buyer questions across five engines with three measured runs each, with crawler access checked first.
What our observations cannot promise
Our dated observations do not promise future naming, first placement or leads. We can define a question set, retain answers and visible sources, and identify supported page corrections. A later matched observation can describe a difference; it does not, by itself, prove that the correction caused it. Assess any separate commercial promise against its written terms and evidence.