Finding an answer when the market data barely existed
How I combined search behavior, community overlap, public data, and marketplace evidence when the usual sources came up short.
Method: Google Trends, audience overlap, community analysis, market sizing, secondary research, web-scraped metadata, and language-model-assisted coding.
What we needed to figure out
The team was looking at new markets and potential partnerships, but some of the opportunities were too small or too specialized to show up in conventional market reports. We still needed a way to compare them and decide where more work would be worthwhile.
What I did
I had to get creative with the data sources. Instead of treating any one number as the answer, I combined several smaller signals, kept the assumptions visible, and used the overall pattern to make recommendations.
How I built the estimate
For geographic and partnership work, I combined Google Trends, current-user benchmarks, community overlap, Reddit audience overlap, and secondary research. That gave me a way to back into estimates of opportunity size and audience fit even when a clean market-size figure did not exist.
For a related marketplace project, I partnered with Engineering on a web scraper and then worked with the data in Python. I used language-model-assisted tagging to make sense of product descriptions and metadata at scale, while auditing the categories and results myself. That let us compare the outside market with our own product and see where the differences actually were.
What the team could do with it
The work gave the team actual numbers to discuss and a clearer view of which markets, partnerships, and marketplace changes deserved attention. Just as importantly, everyone could see where the estimates came from and where the uncertainty remained.