12 of 273 product-truth-evaluable stores had a confirmed public product-truth mismatch.
The publication-grade rate is supporting Breakage evidence. The v1 report does not yet publish a full PDP-vs-schema-vs-sitemap deterministic drift census.
For this report, StoreSteady evaluated a frozen corpus of 500 Shopify storefronts. Freeze: 2026-05-21. Evaluation window: 2026-05-23T16:50:03.820Z to 2026-05-23T23:24:45.338Z.
Headline denominator: 273 stores that passed evidence-bundle and detector eligibility checks within the 500-store corpus.
Run a store-specific scan for PDP copy, schema, canonical URLs, images, price, availability, policy, and checkout evidence. The result is evidence for your store, not a projection from the benchmark.
See whether your store has product-truth drift- Frozen corpus
- 500
- Shopify storefronts entered the public research frame.
- Product-truth evaluable
- 273
- Stores with enough public product evidence to evaluate the Breakage product-truth metric.
- Confirmed mismatch
- 12
- Stores with a calibrated, reproducible product-truth mismatch.
- Access friction separate
- 210
- Stores with access friction are not counted inside the product-truth mismatch rate.
Published metric quality
| Metric | Tier | Evaluable N | Finding N | Flagged N | Non-finding N | Rate | Wilson CI | Quality | Caveat |
|---|---|---|---|---|---|---|---|---|---|
Breakage-confirmed product-truth mismatch product_truth_mismatch | Headline | 273 | 12 | 61 | 139 | 4.4% | 2.53% - 7.52% | alpha 0.9233 / accuracy 99.0% | Publication-grade supporting evidence from the Breakage product_truth_mismatch row; not a connected-merchant-data compliance claim. |
Not in product-truth denominator not_product_truth_evaluable | Limitation | 500 | 227 | - | - | 45.4% | 41.1% - 49.8% | deterministic | Counted outside the product-truth denominator; missing or blocked evidence is not treated as a finding. |
The useful next step is checking whether your own product pages create this same public product-truth drift.
Run a store-specific scan for PDP copy, schema, canonical URLs, images, price, availability, policy, and checkout evidence. The result is evidence for your store, not a projection from the benchmark.
See whether your store has product-truth driftDetector primitives
ProductGroup or variant markup missing
smoke validated; no frozen-corpus rate
Detects visible variants without ProductGroup, hasVariant, productGroupID, or isVariantOf markup.
Offer URL and canonical product URL disagree
smoke validated; no frozen-corpus rate
Compares canonical, sitemap, Offer.url, and selected-variant URL paths.
Product image signals disagree
smoke validated; no frozen-corpus rate
Compares PDP primary image, og:image, and Product.image identity.
Visible and structured prices are ambiguous
smoke validated; no frozen-corpus rate
Detects multiple prices without an obvious sale, member, subscription, or active-price context.
Methodology summary
Selection rule: Frozen external retail Shopify storefront corpus selected before analysis; no merchant-identifying examples are published.
What StoreSteady Measured
The publication-grade row measures product facts that could be compared against public product evidence and that survived calibration and reproducibility gates.
StoreSteady treats this as product-truth drift evidence because the public answerable product fact and observed public product evidence disagree. It is not a connected-feed audit.
- Product-truth mismatch rate: 12 / 273 evaluable stores, Wilson 95% CI 2.53% to 7.52%.
- Judgment quality: alpha 0.9233 and calibration accuracy 98.99%.
- Deterministic PDP/schema drift primitives exist but are not published as corpus-wide rates in v1.
Deterministic Drift Scope
The upstream public-drift detector family covers PDP copy, JSON-LD/Product and ProductGroup markup, Offer.url, canonical URL, sitemap URL, image signals, price text, and availability context.
Those detector primitives were not persisted as a frozen-corpus aggregate in the published Breakage artifact. The report therefore lists them as detector-ready primitives instead of manufacturing a denominator.
What StoreSteady Did Not Measure
The report does not audit private merchant-feed state, does not use private admin evidence, and does not infer hidden model behavior.
It also does not publish named-store examples, raw URLs, exact product names, raw AI responses, source quotes, or private feed/admin evidence.
Quality Gates
The product_truth_mismatch row uses Breakage-level calibration gates: two independent judges, human tiebreak policy, alpha, accuracy, false-positive and false-negative review, and 2-of-3 reproducibility for confirmed candidates.
The deterministic public-surface detector rows require a future frozen-corpus denominator audit before any rate can be published.
Limitations
- The headline is a Breakage-confirmed supporting product-truth row, not a full PDP-vs-schema-vs-sitemap deterministic drift census.
- No private Merchant Center feed, Shopify admin, or connected-store data is used in public examples or claims.
- Dynamic pricing, sale timing, locale, and member-only price contexts can create false positives; future deterministic rates require timestamped crawl windows and tolerance bands.
Artifacts
Artifacts are aggregate-only and include no store names, raw domains, raw URLs, raw crawl text, raw AI responses, exact promo codes, private feed/admin evidence, or merchant-identifying examples.
Correction policy
Methodology gaps, reproducible errors, or source corrections can be sent to research@storesteady.com. Corrections are published with the version history in the methodology page and report JSON.