Metrics we’d put in a board deck.

The brand ranked top-5 for its money keywords, but ChatGPT and Perplexity named three competitors and never them. Diagnosis: an inconsistent entity footprint and zero third-party citations in the sources models retrieved.

We shipped a statistics library their team could maintain, tightened schema and sameAs wiring across 14 profiles, and ran a digital-PR campaign into the trade press LLMs quote most. Citation rate moved within two model refreshes; pipeline followed the referrals.

A question inventory from support tickets and PAA data surfaced 214 real buying questions their pages never answered. We restructured the top 40 category and guide pages: question-led H2s, 50-word direct answers, FAQPage and Product schema.

Six weeks in, snippet ownership jumped; three months in, AI Overviews began citing their buying guides as the source for category comparisons.

Phase one cleared six years of technical debt — crawl waste, duplicate entities, Core Web Vitals in the red. Phase two built topical authority with operations data reporters wanted to cite.

By month nine the brand appeared in more category answers on Perplexity than any competitor, and organic became their cheapest pipeline channel by a factor of four.