Three measurements remain separate
- Storefront Readiness: public crawler access, structured product data, and description readability.
- Product Readiness: completeness of one Shopify product based on Admin API data.
- Observed AI Visibility: what defined providers returned for a saved buyer question at a recorded time.
They are not averaged into one AI ranking score.
Storefront Readiness
| Category | Weight |
|---|---|
| robots.txt crawler access | 20 |
| Product/ProductGroup and Offer JSON-LD | 45 |
| Product-description readability | 35 |
agents.md and llms.txt are recorded but currently contribute no points.
The scanner samples up to 12 discovered product pages. JSON-LD checks seven required groups: name, description, brand, GTIN or MPN, price, currency, and availability. Conflicting Product entities reduce that page’s completeness result. AggregateRating is recorded but is not a bonus.
A description passes when it contains at least 40 words, at least one extractable fact structure or product-fact vocabulary, and is not almost entirely built from generic marketing adjectives.
earned = sum(checkable category ratio × category weight)
available weight = sum(checkable category weight)
score = round(earned / available weight × 100)
coverage = available weight / 100
Partial results must display coverage. A password-protected or unreachable storefront, or a scan with no weighted evidence, returns no score rather than zero.
- 0–59: Your store is hard for AI to understand
- 60–79: A few important gaps
- 80–100: Strong AI shopping foundation
These are SkuWatch readiness bands, not provider ranking bands.
Product Readiness
| Check | Points |
|---|---|
| Product title | 10 |
| Description length | up to 25 |
| Description fact structure | 10 |
| Vendor or brand | 10 |
| Product type or category | 10 |
| Featured image | 10 |
| Featured image alt text | 5 |
| At least one SKU or barcode | 10 |
| At least one valid price | 5 |
| Valid options or a single variant | 5 |
Description length earns 25 points at 40 or more words, 15 at 20–39, 5 at 1–19, and 0 when missing. The maximum is 100. Not checked is not zero.
Observed AI Visibility
Observed AI Visibility keeps provider status, raw answer, citation, entity match, market, prompt version, and time. Failed, timed-out, and excluded provider attempts do not enter the observed mention-rate denominator. It does not generate a universal /100 score.