AI referrals are a real channel
Retail analytics show fast growth in generative-AI referrals, while results still vary materially by site and journey.
For online stores
Can agents read your products, prices, and stock correctly?
Inspect public catalog signals, product structured data, and agent-readable output. This does not execute or verify checkout.
Public HTTP onlyNo login or pluginBounded scan deadline
example.comchecking
robots.txt and sitemappass
Structured datapartial
Agent-readable outputpass
Checking catalog and commerce signalschecking
Diagnostic reportpending
Example surface · real scans show real backend states
Why it matters
Catalog stacks are fragmented
Woo stores commonly combine many plugins, so product facts cannot be assumed correct without checking the rendered result.
A diagnostic is not checkout proof
We inspect prices, availability fields, and public feed signals. We never claim that a real order or payment would succeed.
- [1] Generative AI-powered shopping rises with traffic to retail sites. Adobe Analytics retail data and a US consumer survey; not evidence about every store.
- [2] WooCommerce statistics and benchmarks. An anonymized Metorik dataset; participating stores may differ from all Woo stores.
- [3] Generative AI-powered shopping rises with traffic to retail sites. Adobe Analytics retail data and a US consumer survey; not evidence about every store.
How the scan works
Enter a public URL
No account is required. Syntax is checked before the request is accepted.
Watch real phases
We show queued, running, partial, blocked, and failed states without a fake percent.
Use the diagnostic
The teaser gives a useful baseline. A versioned backend result supplies score and coverage.
What you get back
Example reportCanonical check fixture
No invented score
This preview uses canonical check IDs. A real score, level, coverage, and finding selection only render from the scan status DTO.
Read the scoring methodology →#13 Agent-UA accessibilitypass
#6 Vertical structured-data qualitypartial
#11 OAuth discoveryn/a
Common questions
Do you change anything on my site?
No. We use GET and HEAD to read public HTTP surfaces. We do not log in, run forms, bypass protections, or modify your site.
Do you ask ChatGPT what it says about me?
No. This is an HTTP-readiness diagnostic, not a test of the answers produced by any particular model.
Is the score a certification?
No. It is a versioned diagnostic level that helps orient the findings. It does not promise rankings, traffic, sales, or inclusion in a model.
What if a check cannot complete?
We mark it unavailable. That means we could not reach a verdict; it is not counted as a confirmed failure.