Free tool · Live citation test

Is ChatGPT recommending your product — or your competitor's?

Q: What does each verdict mean?

Cited as a source means ChatGPT retrieved and referenced your page — the strongest signal. Mentioned, not cited means the brand or product appeared in the answer but your PDP was not a source, so the engine is describing you from third-party material. Not surfaced means the answer recommended other products entirely. Each verdict maps to a different fix priority.

ChatGPT shopping answers typically surface one to three products per query. Either yours is in the answer or someone else's is. The checker runs two live buyer prompts through GPT-4o with web search and shows you exactly what the engine said, cited, and skipped.

Free No signup ~60-second result

Report manifest

Two live ChatGPT prompts — category search + buying intent — via GPT-4o with web searchfree

Visibility score and verdict: cited as a source / mentioned, not cited / not surfacedfree

Top 3 findings with the highest-leverage fixfree

Both full ChatGPT responses, verbatim, with every citation linkedemail

Remaining findings ranked by severityemail

Rate limit: 1 run per 30 seconds per IP — each run spends live model and search calls.

How it works

1 · Identify. The checker reads the PDP — title, schema, description — to identify the product, category, and the attributes a buyer would search on.~10s

2 · Prompt. Two buyer-style prompts run live against GPT-4o with web search: a category-search question and a buying-intent question.~40s

3 · Verdict. Responses and citations are parsed: cited as a source, mentioned without citation, or not surfaced — plus ranked findings naming the fix.~10s

What each verdict means — and what fixes it

Verdict	What it tells you	Fix guide
Cited as a source	ChatGPT retrieved your PDP and used it to build the answer. Protect the position: keep schema, price, and availability current so the citation survives refreshes.	Product AI visibility
Mentioned, not cited	The engine knows the brand from third-party material — reviews, Reddit, retailers — but your own page wasn't worth retrieving. Usually a content-depth or schema gap on the PDP.	Rank products in ChatGPT
Not surfaced	The answer recommended other products entirely. Check crawler access first — a blocked OAI-SearchBot makes everything else moot — then entity clarity and answer coverage.	What is GEO

The deeper pattern: ChatGPT answers shopping prompts by synthesizing a handful of retrieved sources. If your PDP cannot be retrieved (crawler access), cannot be parsed (schema), or does not answer the prompt's question (content depth), the slot goes to whoever's page can. The citation analysis glossary entry covers how citations differ from mentions, and the ChatGPT platform page covers how eCommerce Insights tracks this across the catalog.

One engine is a sample

ChatGPT today. Six engines, every product, on the paid plan.

One product on one engine is a useful spot check. The catalog answer needs ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude, and Copilot, re-tested on a schedule, with movement flagged. That is the eCommerce Insights product.

Start free trial See all engines

Related tools

Store-wide

Shopify SKU Visibility Grader

Five sampled PDPs scored for readiness — the breadth companion to this depth check.

Generator

Product schema generator

If the checker says "mentioned, not cited," start here: complete JSON-LD the engine can parse.

Ask AI about the ChatGPT checker

Have your favorite AI engine summarize this tool for your product line.

Frequently asked questions

How does the ChatGPT Product Visibility Checker work?

It reads your product page to identify the product and category, generates buyer-style shopping prompts, runs them live through GPT-4o with web search, and records whether the response cites your domain, mentions the brand without citing the page, or surfaces neither. The full report includes both complete responses and every citation.

Is this a real ChatGPT query or a simulation?

Real. The checker calls GPT-4o with web search at run time — not a cached database or a heuristic guess. That is also why runs are rate-limited to one per 30 seconds per IP: each run spends live model and search calls.

What does each verdict mean?

Cited as a source: ChatGPT retrieved and referenced your page — the strongest signal. Mentioned, not cited: the brand appeared but your PDP was not a source, so the engine is describing you from third-party material. Not surfaced: the answer recommended other products. Each maps to a different fix priority.

My product didn't show up. What do I fix first?

The findings rank it for you, but the usual order is: confirm robots.txt admits OAI-SearchBot and GPTBot, fix Product JSON-LD with the schema generator, then deepen description content. The how to rank products in ChatGPT guide walks the full sequence.

Does the checker test Perplexity or Google AI Overviews too?

No — ChatGPT only, because its shopping answers have the narrowest surface (typically one to three products per query as of mid-2026), which produces the clearest per-product signal. A Perplexity Shopping checker is on the roadmap for Q3 2026. The paid product tracks all six engines on a schedule.

Two prompts free. Every query that matters, tracked.

The checker answers "are we in the answer today?" The product answers it every week, per product, per engine.

Run the checker Start free trial