Funnel Design Library
Funnel design — multi-step strategy. Each playbook is a named funnel play: where the sale actually happens, what every other stage's one job is, and which crawled funnels measurably run it, with the on-page evidence quoted. Read each stack top to bottom — the way a funnel flows: the magenta stage is where the sale happens; everything after it only confirms. The trinity: Funnel designs (the play) → Page designs (the stack) → Section designs (the band). Same blocks, three zooms.
The ad sells; the lander confirms a decision already made.
Too cold for a 30-second sale — an article does the selling.
Search traffic arrives unsold, comparing — candor converts.
The visitor self-segments; the result page is a lander they briefed.
Price outruns impulse — one long page changes the belief.
Paid acquisition buys the first order at a loss; retention is the business.
Clicked, didn’t buy — resume the decision, never restart the pitch.
Method: a funnel matches a playbook only when multiple independent measured signals concur (the confirmation-lander pattern needs 4 of 5 evidence-backed groups: proof mass, authority, risk removal, offer framing, proof above the fold) or one unambiguous quoted signal (a "take the quiz" CTA). Detection is deterministic — the same rules that produce the Evidence Coverage Score; no LLM, no vibes. Matches are SURFACE matches: we claim what's measured on the page, never what a brand's ads or emails do. Plays whose surfaces a crawl can't observe (VSL video, post-purchase email) simply don't produce matches — never guessed. Scope: the 181 funnels we've measured (SaaS-heavy today), auto-updating as the corpus grows. Single-page structure lives in the Page Design Library.