Technical guide · original methodology
How to learn from a private implementation without identifying the business
A confidential pilot methodology using generic lessons, reviewed abstractions, and synthetic examples—not a named case study.
Protect the source, publish the method
Implementation work can improve a research method without making the participating business a public case study. This protocol keeps business identities and raw operational evidence private. Public educational content explains general design choices or uses synthetic examples explicitly labeled as such. It does not present a disguised business profile as independent validation.
Remove more than the name
A domain, logo, screenshot, product name, exact revenue figure, unusual program, distinctive route, timestamp, or combination of details can identify a business. Removing the title of a case study is therefore insufficient. The public example must be evaluated for indirect identification, including metadata, downloadable files, source citations, structured data, machine manifests, and social copy.
Preserve a private working record
An authorized working record may retain the evidence needed for debugging and verification. It belongs in access-controlled storage outside public APIs and build assets. Public schemas must not serialize private site keys, domain hashes, source maps, or evidence links. Synthetic fixtures use reserved example domains and invented values rather than transformed customer records.
Distinguish method from outcome
A generic implementation lesson can be useful without publishing a private performance metric. It must not be framed as a measured industry result. Operator-managed work is not independent evidence, and an anonymized single-site example is not a representative cohort. Any future approved result needs a scope, methodology, uncertainty statement, and separate release review.
Default to no public case study
The default release contains methods, checklists, and synthetic examples, not pilot records. Publishing a business identity, identifiable evidence, or an actual pilot case study requires separate explicit authorization. Social distribution, agent-readable resources, and partner-site content follow the same rule as the human article.
Sources and scope
AIWebSignals: current public capability contract · Google: intrusive and mandatory interstitials
Provider documents support the referenced technical facts. The interpretation, examples and proposed checks are AIWebSignals methodology, not provider endorsement or measured industry results.
Review the assessment method and its limits · Request a correction privately · Read the JSON version