An SEO engine that tripled qualified leads in two quarters.

Aether sold mobility subscriptions across fragmented city pairs. Marketing celebrated impressions; finance asked why CAC from organic barely moved. The site stack was a Next.js marketing layer bolted onto a blog nobody indexed, while product pages duplicated parameters Google treated as soft duplicates.

The trigger was a board request to show compounding acquisition without buying another paid month. Legal also wanted clearer regional copy for pricing — which meant we could not "SEO" our way out with gray tricks. Timeline: two quarters of runway before the next raise story had to land.

We inherited brittle CMS permissions, a sitemap that listed URLs that 404ed, and a reporting layer that double-counted returning users as new trials. Eleni led the engagement; we agreed upfront that rankings were a means, not the success metric.

// TODO: Verify with client legal before publishing — market list, funnel definitions, and traffic multiples.

Framing. This was demand capture engineering: crawl budget, internal links, and template hygiene were product features. IA. We collapsed faceted URLs into canonical patterns and introduced a hub-and-spoke model for city pairs so authority pooled instead of diluting.

Phasing. Month one: technical baseline, render logs, and CWV fixes on templates that carried money keywords. Month two: programmatic expansion with human QA gates — no page shipped without a risk label. Month three onward: backlinks only where editorial fit existed; we refused paid link farms the CEO’s cousin recommended.

Team. Eleni paired with Aether’s head of growth; a Veltris engineer owned edge caching and edge middleware for hreflang. Judgment. We blocked a 600-page "dictionary" proposal that would have polluted topical relevance; we shipped 220 pages instead, each tied to a service line.

// TODO: Verify with client legal before publishing — vendor tooling, page counts, and experiment windows.

Qualified leads — using Aether’s tightened SQL definition — rose 3.1x in two quarters while organic sessions rose 218% on the same landing inventory (expanded, but measured apples-to-apples on overlapping URLs). Forty-two priority terms reached #1 in target locales where they previously oscillated on page two.

Product finally trusted the SEO roadmap because experiments tied to revenue sat in the same spreadsheet as technical tickets. Paid spend dialled down modestly without a pipeline cliff — the outcome the CFO cared about.

We continue on a quarterly advisory retainer: CWV regressions, indexation alerts, and content experiments. The engagement closed its heavy build phase cleanly with documentation their new hire extended.

// TODO: Verify with client legal before publishing — revenue and paid media interaction claims.

We underestimated how long sales would hold onto a deprecated SQL definition — it delayed celebration charts by six weeks. We were surprised how much uplift came from fixing parameter handling vs. new copy; it reinforced that crawl hygiene is a creative problem.

We would start with a single executive dashboard wired to CRM stages on day one, not month three. Methodology-wise we now require a written "kill list" for content ideas before research hours burn.

// TODO: Verify with client legal before publishing — internal political details anonymized but still sensitive.