I ran how the marketplace surfaced and sold its catalog
620 · category pages I launched
Outschool had thousands of classes and no deliberate system for putting the right ones in front of the right parents. Discovery was thin, SEO was weak, and there was no way to build or test a page without engineering.
The problem
We were leaving demand on the table. Parents searched in patterns we weren't acting on, our best inventory was hard to find, and every new marketing page meant an eng and design cycle.
What I owned
I built the merchandising layer: our first real catalog taxonomy (by hand, in Excel, pre-LLM) that fed search tagging and the navbar, plus suggested searches, a seasonal global nav, and the reading of demand trends that told product and marketing what to push and when.
Turning that into pages
The team started generating category landing pages at scale to get indexed, about 25,000 of them, with early ChatGPT drafting hero copy that I edited and maintained. That SEO project turned into a self-serve system (headless CMS) that let marketing ship and test pages in minutes instead of weeks, with no eng/design loop. I helped shape it and then it became my surface to run.
What moved
These pages became a core part of the organic and paid machine. I launched 620 by hand: wrote the copy, chose the imagery, ran the tests, and experimented with early LLMs to generate copy and images. Together they drove hundreds of thousands in revenue. I also moved our catalog ads onto category pages with no supply limits, which lifted paid efficiency without cutting conversion. The real unlock was cadence. Marketing stopped waiting on the backlog, and I could tune what the marketplace merchandised as fast as demand moved.