$ man find-similar

Find Similar

Exa SDK function that takes a seed URL and returns companies with similar web presence, positioning, and content. TAM expansion in one API call — give it your best customers, get back companies that look like them.

by Shawn Tenam

building a target account list usually means manual research — searching LinkedIn, scanning industry directories, asking for referrals. find_similar() flips that. you give it your 5-10 best-fit companies and it finds more like them. it's not keyword matching — it's semantic similarity based on web content. a supply chain consulting firm and a 3PL strategy firm look different on paper but similar in Exa's model. that's the kind of match you'd miss with traditional search. I used it to expand a partner's TAM by 19 net-new companies from 10 seed URLs. those are companies we never would have found manually.

I pick 5-10 diverse seed companies from the top tier of an existing account list — different segments, different sizes, but all strong ICP fits. I pass their domains to exa.find_similar() with num_results=20 (wide net). the response includes company descriptions, domain authority signals, and content summaries. I then dedup against the existing account list, remove known companies, and score the remainder against ICP criteria. the output is a CSV of net-new lookalike companies ready for qualification. the key insight: diverse seeds produce better results than similar seeds. if all your seeds are the same type of company, find_similar() returns the same type. spread your seeds across segments and you discover adjacent markets you didn't know existed.