Evidence-first diagnostics

How CommerceLint audits a store.

The objective is not to award a badge. It is to produce a reproducible repair backlog in which every recommendation is tied to observed commerce evidence.

1. Establish the purchasable truth

CommerceLint begins with what a shopper can actually select and buy: product identity, selected variant, current price, currency, availability, URL state, shipping terms, and return conditions. Structured data, feeds, page copy, and checkout context are compared with that customer-visible state.

Priority rule: an incorrect purchase fact is more urgent than a missing optional field.

2. Sample representative catalog states

A useful audit does not inspect only the cleanest product. The founding audit selects representative states such as an in-stock item, out-of-stock item, discounted product, variable product, product with multiple images, and any template or category known to behave differently.

The sample is intended to expose systemic template and integration defects. It is not presented as a complete crawl of every product unless that scope is separately agreed.

3. Preserve reproducible evidence

Each finding should identify the affected URL, observed value, expected value, source surface, probable system owner, customer or distribution risk, recommended repair, and verification method. Raw evidence is favored over unexplained pass/fail labels.

Evidence fieldPurpose
Affected URLLets the team reproduce the finding.
Observed valueRecords what the page, schema, feed, or policy currently says.
Expected valueDefines the storefront truth or required correction.
PrioritySeparates purchase-risk defects from completeness opportunities.
VerificationDefines how the team will prove the repair worked.

4. Score by commercial risk

The free scanner weights missing Product and Offer fields more heavily than lower-risk completeness fields; it does not claim that a present value is accurate. The paid audit verifies contradictions and uses any score only as a summary, not as the deliverable. Incorrect price, currency, stock state, or variant identity should dominate cosmetic or optional omissions.

A score is never treated as proof that a search engine or shopping agent will index, recommend, or purchase the item.

5. Create a repair order

  1. Correct unsafe or contradictory purchase facts.
  2. Repair product identity, identifiers, canonical state, and variant relationships.
  3. Synchronize shipping, returns, feeds, and storefront surfaces.
  4. Add useful completeness fields only after the underlying facts agree.
  5. Run regression checks across the representative catalog states.

6. State limitations plainly

CommerceLint performs technical diagnostics on public material supplied or selected for review. Platform requirements and product-discovery systems can change. Results must be validated against the live storefront and current primary documentation. CommerceLint does not guarantee rankings, inclusion, recommendations, purchases, traffic, or revenue.

Use the method on your store.

Start with the free local scanner, review the sample report, or request the founding audit for a prioritized implementation backlog.