249 sponsor prospects for $2.87 in API spend

The problem: API spend with no ceiling

Doing it by hand meant hours in Google Maps. Doing it naively meant paying per scraped result for a list padded with popular but irrelevant places, because map search results optimize for popularity, not for fit. The job needed both a cost ceiling and a relevance filter that could not be gamed by the search engine's own padding.

The approach

A five-stage re-runnable pipeline: build the search space, pull business data through a scraping API under an explicit dollar cap passed in at runtime, enrich each record with contact and social details from its own website for free, then classify, deduplicate and prioritize.

The classification rule is the part that mattered: businesses are categorized on their own name and category, never on the search term that surfaced them. That is what keeps padded irrelevant results out of the final list.

Result

• 249 prioritized businesses delivered with contact and social details attached.
• Total API spend of $2.87 against a $5 ceiling: roughly $0.0115 per delivered business.
• Cost bounded at runtime by a cap argument, so an expanded run cannot surprise anyone with a bill.
• Fully re-runnable: new cities or categories are a config change, not a rebuild.

Related, but a different ceiling: the cold email engine bounds sending volume while working a list. This pipeline bounds API spend while building one.

Built by Hisham Hanawi. This is a systems integration build.