What to do when ChatGPT describes your brand wrong
Being described incorrectly is a bigger commercial problem than being absent, and it is invisible unless someone reads the answers. You cannot correct an assistant directly. The fix is changing the sources it retrieves, which means finding the specific page carrying the error and dealing with that.
Why does an assistant get my product wrong?
Because it is repeating a source. Almost always an outdated third-party page, an old review, or a comparison written before your last release.
That is good news, because it makes the problem tractable. A model inventing something from nothing is hard to address. A model faithfully repeating a two-year-old roundup is a page you can go and get updated.
How to fix it, in order
Read the answers containing the error and look at what they cite. The wrong fact usually traces to one or two widely cited pages.
Most publishers update a factual error when asked with evidence. This is unglamorous outreach and the single highest-yield action.
A clear, dated page on your own domain giving the current price, feature or certification. Vendor pages do get cited for factual sub-questions.
Answers vary run to run, so confirm the correction with repeated runs before declaring it fixed.
How long does a correction take to show up?
It depends on retrieval rather than on you. Live-retrieval answers can change within days of the source updating; anything leaning on training data takes far longer or never changes.
Common questions
There is no reliable per-brand correction channel. Changing the sources is the mechanism that works.
In regulated categories, yes. A wrong claim about fees, eligibility or certification is a compliance exposure as well as a lost sale.
Writes up what the stored answers show, one category at a time.