B2B market research API

Sizing a market

company-search with an industry keyword, a geoId and a companySize filter gives you a count and a pageable list. Run the same query across size bands and you have a distribution rather than a single number — which is usually the question anyway, since "how many companies" matters far less than "how many of the size we can sell to".

Store the query and the date. Re-run it quarterly and the delta is the finding: a segment that grew 12% in six months is a different investment case from one that is flat, and no analyst report will tell you that about a niche.

Hiring as a growth proxy

Companies hire ahead of revenue and stop hiring ahead of trouble, and both are public. job-search filtered by companyId and role keywords, sampled weekly, gives you a growth signal for a whole segment for a few dollars a month — it is the cheapest endpoint here at $0.0100.

Aggregate rather than reading individual posts. "Postings for this role across our 200 target accounts are up 40% quarter on quarter" is a finding. One company posting one job is noise.

Listening to the practitioners

post-search on the vocabulary of a problem tells you how people describe it when they are not talking to a vendor. That is genuinely hard to get any other way — survey answers are shaped by the question, and sales calls are shaped by the seller.

group-search finds where a niche audience gathers, and post search scoped to a group reads the conversation inside it. For an unfashionable corner of B2B, that is often the only qualitative source available, and a few dollars of balance buys a real read on what a professional community is arguing about this quarter.

Being honest about the bias

This is a sample of people who are on LinkedIn and choose to post there. That skews heavily toward white-collar roles, English-speaking markets, larger companies and people with a professional reason to be visible. A finding about "the market" drawn from it is really a finding about that slice.

It is also self-reported. Employee counts come from how many members list a company as their employer, not from payroll. Job posts sometimes stay live after a role is filled. Treat everything here as a strong relative signal and a weak absolute one — track the direction and the delta, and be careful about quoting the level.

Sampling on a fixed cadence

The value here compounds only if the query and the schedule stay fixed. Change either and the series before and after are not comparable, which quietly destroys the trend you were building.

Pick a cadence you can sustain, store the exact parameters with each run, and resist improving the query mid-series. If it must change, start a new series alongside the old one rather than replacing it.