StoreSteadyStoreSteady
Public rubric v1.0

AI Commerce QA Rubric

The rubric explains the public gates a StoreSteady finding should pass before it is useful to a merchant.

Public gates

Evidence anchor

The finding must point to public evidence, merchant-connected evidence, or an observed AI or channel response.

Specific commerce consequence

The finding should explain the merchant-facing consequence without inflating it into a ranking, revenue, or compliance promise.

Bounded recommendation

The recommendation should name the source surface to fix first, such as a PDP, policy page, structured field, feed value, or checkout step.

Confidence label

The output should say whether the issue is confirmed, likely, needs merchant confirmation, or inconclusive.

Validation path

The finding should explain how StoreSteady or the merchant can retest the same source after the fix.

What fails the rubric

Unsupported certainty

Claims like guaranteed rankings, guaranteed AI visibility, or guaranteed revenue outcomes fail unless scoped evidence supports them.

Generic SaaS advice

Advice that could apply to any store without naming the evidence source does not qualify as an AI Commerce QA finding.

Private-system assumptions

StoreSteady does not claim what happens inside private tools, catalogs, or channel accounts unless the merchant connects that source.

Confidence handling

Confidence is not a decoration. It controls how strongly the finding is worded, whether merchant confirmation is needed, and which validation path is appropriate.

Validation path

A fix queue item should include the retest: re-crawl the page, re-read structured data, compare feed values, repeat the AI or channel observation, or rerun the supported checkout probe.