Research standard

Evidence before software.

Our work separates search demand, product evaluation, and product-building decisions. None of those signals is allowed to substitute for the others.

1. Keyword qualification

We prioritize stable English search terms with commercial or problem-solving intent. For a new domain, the preferred starting range is roughly 100-3,000 US searches per month and KD below 30.

2. SERP and content quality

Before a non-brand keyword enters the publishing queue, its result page must show an opening: independent sites, mixed page types, outdated material, weak comparison structure, or questions current pages do not answer.

3. Controlled product evaluation

Alternatives and comparison articles use the same representative workflow across candidates. We measure time to a finished outcome, correction burden, integrations, operating cost, and failure modes.

4. Commercial disclosure

AI Marketing Atlas may eventually build a product in a category we research. Product hypotheses are labelled as labs, and interest capture is separated from editorial conclusions.

5. The build gate

A category advances only after ranking, traffic, qualified waitlist, conversion, interview, and buildability thresholds are all met.

6. Corrections and maintenance

Commercial pages include verification dates. Material pricing or product changes trigger review.