2,400 pages in six weeks, and not one of them embarrassing.

The brand had been growing 40% YoY for three years. The team scaled operations, fulfillment, and paid media — but content didn't keep up. Of 600 active SKUs, roughly 200 had finished product pages with real copy. The rest had a name, a photo, and a 30-word stub. Category pages didn't exist — the site structure was category landing pages that routed straight to grids. Google had nothing to rank.

The two-person content team could write two solid product pages a day. At that pace, finishing the catalog alone was 10 months of work — while the brand kept launching new products every week. The math didn't work.

Two weeks of setup, four weeks of production. The engine had five pieces:

• Product data normalizer. Pulled from Shopify, plus the brand's lab-test PDFs, supplier spec sheets, and internal knowledge wiki. Structured every SKU into a consistent data object: what it's made of, who it's for, how to use it, what problem it solves.
• Voice-trained drafter. Week one was spent extracting the brand's voice from the 200 good existing pages. Principles went into a style guide the model used on every draft. Output read like the brand — not like ChatGPT default.
• Quality gate. Every draft ran through automated checks (factual grounding in the source data, banned phrases, reading level, keyword inclusion) plus a human review queue. Below-threshold drafts bounced back for another pass. Nothing shipped that a human hadn't actively approved.
• Category page builder. Category pages weren't just product grids. The engine generated intro copy ("what to look for," "how we test"), comparison tables, FAQs, and buying guides — real content, driven by real data.
• Programmatic SEO layer. On top of core product and category pages, the engine generated targeted landing pages for long-tail queries ("best X for Y," "X vs Y comparison"). Each one grounded in actual product data, not fabricated claims.

The two content people didn't write a single product page during the engagement. They did something more valuable: they wrote the voice guide, reviewed the first 100 drafts until the output felt right, and spent the rest of their time building the brand's editorial properties — the seasonal gift guides, the founder story, the deep-dive blog posts that actually convert.

When the engine finished its run, the team had 2,400 quality pages live and a real editorial backlog they were shipping against — not a copy-writing backlog they'd never finish.

In the quarter following launch, organic traffic grew 340% and organic revenue grew 68%. (Revenue grew slower than traffic because a lot of the new traffic was top-of-funnel buying-guide queries — still valuable long-term, but they convert at a lower rate than product-name queries.) The brand's programmatic SEO pages now rank in the top 10 for ~800 commercial queries they had zero presence on before.

Meanwhile the content team — still two people — now ships editorial weekly and has the highest-engagement brand blog in their category. The content bottleneck is gone.