Every engagement ships three layers: the deck your leadership sees, the report that holds the complete reasoning, and the working papers, where the method earns its keep. When your board asks how you know, you hand them the answer instead of defending a slide.
A recommendation means nothing until you know what it beat.
Killed in verification. The proof to back "AI-native" does not exist yet, the claim is saturated across the category, and the modern analogs that tried it without proof are the cautionary shelf.
Killed in verification. On the incumbent's own front the subject is indistinguishable from a rival thirty times its size, and the claim requires proof assets only the incumbent holds.
Positioning pays back in quarters, not on launch day. So every claim the chosen position needs but cannot yet back gets a banking artifact, a time box, and a kill criterion that tells you to fall back.
The layer that justifies the price. A champion who can answer "how do you know?" with the full verifier verdicts and research provenance is a champion who wins the internal argument.
Claim tested: "the only close platform built for multi-entity retail." Verifier verdict: survives, with a raise. The wedge is defensible and the demand is verified in win/loss, but the believability score was raised ⬆ because the two reference customers are both pre-2024 and neither is multi-entity.
Owned-word test: "close" is contested by the incumbent's ad spend; "entity" is unclaimed across the field. Score rationale and the full transcript of the adversarial pass follow.
Once a month I run the full engine on a public B2B company at a visible positioning moment and publish the working: the divergences, the scorecard, the bets, and the grid of what died. The client version gets the call. Client engagements stay private until a founding client trades a named case study, and then it appears here.