Recommendation Audit
Paste a Shopify product link. Colter tells you how to improve the page. Free and read-only.
TL;DR: Run the Recommendation Audit with one public Shopify product page. Colter tells you what to fix first, then lets you check the same page again.
Recommendation Audit is Colter's merchant product. Check, Fix, Test, Lens, and Verify remain available as technical capabilities inside the audit and through the CLI, API, SDK, and MCP server.
The Four-Stage Loop
| Stage | Question | Colter capability |
|---|---|---|
| Baseline | What public evidence is present on this product page? | Check and recommend-audit |
| Fix | What single evidence gap is most likely to make the answer less accurate or complete? | Fix guidance |
| Prove | What did the same named engine actually answer before and after the correction? | Content-hashed answer observations and, when needed, Test |
| Monitor | Did the unchanged prompt gain, retain, lose, or still omit the product or target citation? | Comparable reruns and, when useful, Lens |
What You Need
- one exact public Shopify product-page link
The audit is read-only. It does not change Shopify, publish product content, scrape private answer history, or ask for a model credential.
Web Workflow
- Open Recommendation Audit.
- Enter the product page link. A homepage handoff starts the check immediately.
- If product details are missing, review one suggested correction and copy it for whoever edits the page.
- Apply the correction outside Colter only after review.
- Check the same page again. Colter reports whether the missing page evidence changed.
If the page checks pass, try a shopping question in your usual AI assistant and save the answer outside Colter. If the page cannot be read, check that the link is public before editing any product details. The web audit does not import or compare AI answers, and a verified page detail does not prove a recommendation.
Colter also records five deterministic buyer questions derived from the product name, brand, and category. Those questions are for later repeat testing and do not affect the page findings or fix selection.
Every completed web audit gets an opaque saved-report URL and a content-hashed
evidence receipt. Saved reports are private-by-link and noindex; Colter does
not publish a merchant audit into search results without a separate publishing
decision.
Saved reports show when the page was checked. You can download the receipt or start a fresh check of the same product; that fresh check does not compare against the saved report.
CLI Workflow
colter recommend-audit https://store.example.com \
--product https://store.example.com/products/example \
--intent "best lightweight travel backpack under $150" \
--json
Without an observed-answer file, the command reports a readiness baseline only. A readiness score is not proof of placement, a mention is not a recommendation, and a changed score is not proof that an answer changed.
Evidence Boundary
Colter keeps three classes of output separate:
- Measured: public storefront facts retrieved by Colter.
- Observed: merchant-domain references and citations present in an imported answer from one named engine and prompt; product identity only when manually verified.
- Inferred: a correction priority or mechanism that must still be tested.
Success is a comparable answer delta: gained, retained, lost, or still absent. The audit does not promise that a correction will cause a model to recommend a product.
Technical Capabilities
- Check creates the public-evidence baseline.
- Fix turns an evidenced gap into a bounded correction.
- Test answers a runtime question only when public evidence is insufficient.
- Lens monitors observed agent traffic when ongoing measurement is warranted.
- Verify produces conformance evidence for technical teams.