The challenge

Fifty-eight expertise pages averaged 900 words with five FAQs each; 52 of them had no meta title so Google was inventing its own. One site had 125 URLs Google refused to index — 59 of them near-duplicate blog posts. Schema was emitting a duplicate LocalBusiness on every page. Four domains ran four separate analytics properties with no cross-domain linking, so a visitor moving between sites became four visitors. And AI assistants weren't citing any of it.

What we built

• Expertise library rewrite — 58 pages — Each page rewritten and checked against primary legal sources, applying the partners' written answers. 52,289 words became 133,052 (×2.5); FAQ items went from 294 to 617; 38 H1s rewritten; every page got a real title tag. Staged on -v2 preview URLs for approval, side-by-side with the live version.
• Indexing clean-up — 125 unindexed URLs triaged into utility pages, 27 duplicate-topic clusters, thin pillar pages and templated city pages — with a decision sheet (keep / redirect / delete) the partners could fill in 15 minutes.
• Schema and E-E-A-T — LegalService, LocalBusiness, Person (both partners, with LinkedIn and bar-association memberOf), FAQPage, Article with images, Review/aggregateRating on the PRO tools — deduplicated and green in the Rich Results Test.
• Cross-domain attribution — A portfolio GA4 property deployed through all four GTM containers alongside the per-site properties, with the linker parameter verified on every cross-domain click — paid, organic, direct and email in one view.
• Conversion tracking on PRO tools — tool_start, tool_submit and CTA events on the paid calculators, named to mirror the existing free-tool events so reports stay comparable.
• GEO — generative engine optimisation — llms.txt and llms-full.txt, AI-crawler directives, an /ai-info page, IndexNow and Bing submission, and a content cluster pairing each pillar blog with its PRO tool and expertise page, linked both ways.

How we worked

Nothing published without sign-off: rewrites went to preview URLs with a before/after spreadsheet listing every substantive correction. Legal wording that had to come from a licensed lawyer was flagged as such rather than drafted by us. Weekly written updates with a quick-wins scoreboard.

Outcome

A legal content library two and a half times deeper with every page properly titled, a single analytics view across the whole portfolio, clean structured data that passes validation on every template, and the site plumbing (llms.txt, author entities, topic clusters) that makes the firms citable by AI assistants — all shipped alongside the partners' day-to-day practice.