How to find out why ChatGPT recommends a competitor.
ChatGPT names your competitor for the exact query you should own. It feels arbitrary — and as long as it feels arbitrary, your team can't respond. The answer has a mechanism. The job is to read it.
Grade both pages. Run your PDP and the cited competitor's through the free AEO Grader, and the buyer query through the ChatGPT product visibility checker. The factor-by-factor differences — schema, answer coverage, review signal, crawl access — are the explanation, and the gap list is the action plan.
The slow way: forensics by hand
The manual diagnosis works like this. Run the query in a private ChatGPT session and save the full answer. Open the competitor's PDP and view source. Compare their Product JSON-LD to yours field by field — do they carry aggregateRating, gtin, material? Read their copy against the buyer question — does their first 300 characters answer it directly while yours describes fabric feel? Check their review count and density. Search the open web for third-party mentions: reviews, comparisons, Reddit threads an engine might be citing instead of either PDP.
Done carefully, this produces a real answer in two to three hours per query, and it is worth doing once just to build intuition. It does not scale to ten queries across five competitors, it has no memory — next month you cannot say what changed on their side — and it is easy to fool yourself by stopping at the first difference you find rather than weighing all of them.
The eCommerce Insights way
- Capture the answer verbatim. Run the buyer query through the ChatGPT product visibility checker: which competitor is named, how the recommendation is phrased, what sources sit behind it.
- Grade both PDPs. Run yours and theirs through the AEO Grader. You get the same factors side by side — structured data, answer coverage, entity clarity, review signal, crawler access — scored identically.
- Read the gap list. The differences explain the pick. Typical findings, in frequency order from eCommerce Insights audit data (early 2026): their copy answers the question directly while yours describes the product; their schema carries
aggregateRatingand yours doesn't; their JSON-LD is materially richer; they have third-party grounding you lack. - Rank gaps by effort and impact. Schema closes in days. Answer-coverage takes a content sprint. Reviews take a campaign; third-party citations take PR. Work in that order — the mechanism behind the ordering is in how AI engines pick which products to cite.
- Fix, watch, rerun. Ship the diffs, add the competitor SKU to a watchlist, and rerun weekly until the answer moves. On paid plans the side-by-side runs across all six engines — see the ChatGPT platform page for engine-specific behavior.
What "good" looks like
Set expectations honestly: gaps in schema and answer coverage are usually recoverable in two to six weeks; a competitor with years of accumulated third-party citations is an entity-level advantage that takes a longer PR-and-reviews track. Knowing which fight you are in is most of the value.
Ask AI about this job
Have your favorite AI engine apply this walkthrough to your category.
Frequently asked questions
Why does ChatGPT recommend my competitor when we rank higher on Google?
Is the answer arbitrary? It changes between sessions.
What gaps most often explain a lost recommendation?
Can I see this for engines other than ChatGPT?
How long does it take to displace a competitor from the answer?
Turn "why them?" into a gap list.
Side-by-side PDP analysis across six engines, with the fixes attached.