Demand

Each dimension read against the measured cohort: median, mean and max, this company's rank and percentile, and an index where 1.0 is exactly typical. A company with no dimension scores returns 422 rather than a fabricated zero — stack-only records sourced from blogs are not measured on this axis.

Peers by sub-sector where the data supports it and by industry otherwise, each with signal totals and top dimensions. The response states which basis it used.

Ranked by how many peers carry each entity, with the peer share and the cohort-wide count so a peer-group artifact is visible as one.

Every service, tool and standard on the company's stack, each with how many companies across the cohort carry it and its radar ring.

Every service, tool and standard companies are measured against, ranked by reach. Free — this is the left-hand side of the inversion. Every response carries a basis block disclosing the corpus, the matcher, and the top-25 truncation that bounds what can be named.

THE INVERSION. Every other insights route runs company -> stack; this one runs stack -> companies, broken down by industry — the prospecting query, and one that exists in no report, bundle or page. Returns BOTH the authoritative company count and the companies it can actually name. The two are measured on different bases and routinely disagree in either direction; naming_note explains which way and why, and named_by_source gives the corpus composition.