Free AI accuracy scan
Is AI wrong about your brand?
Roughly 1 in 10 AI answers to a factual question is wrong (read the AI accuracy research report): closed when they're open, wrong prices, wrong policies. Pick one fact about your business and see what Perplexity is actually telling shoppers. Free, 30 seconds.
What the free scan checks
- Whether Perplexity's answer about the one fact you give (business status, hours, a price, a policy) matches the correct value you enter.
- Three differently phrased versions of the question, so one lucky or unlucky answer does not decide the verdict.
- A judged verdict for every answer: consistent with your fact, contradicts it, or does not address it.
- A SHA-256 hash and UTC timestamp for every captured answer, so the record can be verified later.
How the scan works
- Tell us one fact. Enter your brand name, website, the kind of fact to check, and its correct value. That value is the ground truth every answer is judged against.
- We ask Perplexity three ways. The scan sends three differently phrased questions about that fact to Perplexity and captures each answer with a SHA-256 hash and timestamp.
- A judge compares each answer to your fact. An LLM judge classifies each answer as consistent, contradicting, or not addressing your fact. You see the contradiction count immediately; the per-answer breakdown unlocks with your email.
What you get
Immediately: how many of the three answers contradicted your fact. After you enter your email: every answer in full, the judged verdict and the claim it was judged on, and the SHA-256 hash and timestamp for each capture. The result is a one-off spot check; weekly monitoring across five engines is what the paid Accuracy Alerts add.
Questions about the free scan
Is it really free?
Yes. The scan runs without an account, a card, or an email address. Only the per-answer breakdown asks for your email, and you can stop at the summary if you prefer.
Which AI engines does it use?
The free scan asks Perplexity: three phrasings of your question, one answer each. The paid Accuracy Alerts monitor in the PositionBird Shopify app runs the same check across ChatGPT, Perplexity, Gemini, Claude, and Google AI results, nine samples per fact every week.
Is this what shoppers actually see?
Close, but not identical. We query Perplexity through its API (the sonar model) rather than the consumer app, so wording can differ from what a shopper sees in the app, and AI answers vary from one ask to the next. Treat the scan as a spot check, not a verdict.
What happens to my email and results?
If you unlock the breakdown, we store your email with the scan results so we can send them to you and tell you when PositionBird launches on the Shopify App Store. We record the exact consent wording and the time you gave it, we do not sell or share the address, and we delete it on request, as described in our privacy policy.
Full details on data handling are in our privacy policy.