Execution is observable. So is whether anyone cared.

Public sources only — company sites and changelogs, public repos, job boards, Hacker News and Bluesky. Each observation is stored with its URL and timestamp, a move exists only where two consecutive observations differ, and each reaction is bound to a specific dated launch rather than counted loosely against a company name.

By exact-phrase and domain-scoped matching within the window around the launch date. Loose name matching is worthless at this job — searching a bare company name returned six figures of unrelated results in our own testing. The precision of that coupling is the part we spend the most effort on.

Not self-serve yet. Sector coverage expands based on demand rather than speculatively, so the honest answer today is to tell us the sector and we'll say whether it's on the list. We'd rather say no than publish a thin report.

Closer to diligence. It answers whether a company's shipping actually earns a reaction from its market, and how that compares to the rest of its sector. It does not rank companies for you or predict outcomes.

Every published figure is reproducible from the cited sources, and the reports page carries the same totals you can add up yourself. Where coverage is thin for a company we label it rather than smoothing over it.