Company Data API: Search, Enrich, Write Back

What a company data API is
A company data API gives you a business record through a call. The useful fields are the ones a product can store without a second cleanup job: legal or trade name, domain, industry, employee count or band, location, and identifiers you can join on later. Funding, workforce history, and technologies show up in the #1 page's definition. They are optional blocks, not the identity of the company.
One company, one id. Subsidiaries are separate ids with a parent pointer. If you flatten them into a single "company name" string, every enrich after that is a guess. Entity resolution is the work of deciding that two sources are the same firm. The #1 page is right to put that in the opening. It is also the part partners skip when they match on domain alone and call it done. Domain match is a start. It is wrong for companies that share a domain, and for companies that changed domains.
The parent topic in the 30 Sep 2026 export is business data API, at a much smaller volume than this page. Do not quote that volume as this page's demand. This URL owns company data API. People are not a company field. They live on people data API.
Multi-source records you can search
Natural language and semantic search are convenient for a human in a playground. They are a poor contract for a product. A partner's server should send explicit filters: country, industry, employee band, domain. Keep a free-text query if you must, and log the parsed filters you actually ran. Otherwise you cannot explain a result.
Historical data (headcount last year versus now) is a series, not a single integer called employees. If you only store the latest, say so. The #1 FAQ treats historical headcount as a question people ask. Answer it in the schema. Do not imply a time series you do not keep.
Fresh means a confirmed_at on the fields that rot: headcount, domain, and status (operating or not). A record from 2016 with a new export date is not fresh.
Try it before a sales call
Step one. A playground with a known domain and a domain you expect to miss. If the miss still returns a lookalike company, leave.
Step two. A key you can mint yourself, with a scope that can read company fields and cannot read personal phone. The Zoho-style lesson from other queries applies here: scope is part of the API.
Step three. Retrieve the full record by id, not only the search snippet. Search results are thin on purpose. The enrich call is the fat one. Document both URLs. The #1 page shows Elasticsearch-style search paths. You do not need that syntax. You need a search call and a get-by-id call that a partner can copy.
Use cases that are actually different
The #1 page repeats enrich twice and adds AI applications, ICP search, lookalikes, and market research. Separate them or the API becomes one bloated route.
Power a product or an agent. The caller has a domain from a signup and needs industry and size to route the account. That is enrich. Latency matters. A miss must be fast and explicit.
Find companies that match a profile. Filters, a page size, a cursor. This is search. It returns a population nobody has a relationship with yet. Do not pipe it to a sequencer inside the same response.
Enrich an account you already work. Fill nulls on a CRM row. Do not overwrite a value a human confirmed unless the API's confirmed_at is newer and you have a rule for that.
Lookalikes. A search seeded by one company. Label it as a suggestion. It is not a statement that the neighbor is a buyer.
Market research. Aggregates and history. Different rate limit, different cache, often a different buyer. If you only have per-record enrich, do not pretend you have a research product. The #1 page's note on accelerating scientific research is a pro bono program, not a use case for go-to-market. Mention it only to say this URL is not that program.
Delivery, proof, and price shape
Proof on that page is a customer count and a founding year. Do not answer with a made-up count. Proof a partner can use is a domain they name, returned correctly, with a date. Offer that in the playground. A logo wall does not debug a bad join.
Price shape, without our number. Say whether search and enrich draw from the same unit. Say whether a miss is free. Say whether the trial key keeps working after the trial, which is a point their page makes, so people are not surprised either way. Then stop. A partner who needs a commercial conversation can book a call.
Owned records, verified reach, and why now
Search is not an owned record. Enrich is. Start from the account id in the partner's product. Attach technologies through technographic data API only after the company id is stable.
Verified reach does not live on the company. Companies do not have inboxes you should mail by default. A role address on a domain is a hypothesis. Run it through email verification API and still prefer a person. One-off human checks use the email verifier.
Why now is intent on the account, from intent data and buyer intent data, after the company is resolved. A funding round on the record can be a fact. It is not automatically a reason to contact everyone at the domain.
Into the tool that already sends
Activate stays in the partner's system. The company API's last step is a write of the company id and the fields they asked for. Identity resolution then attaches people. The wider practice is data enrichment and B2B data enrichment. The buy-versus-build framing is B2B data provider. Identity resolution software is the system. Identity resolution API is the call. This page is only the company payload.
If the only way to use the record is a UI we operate, we have built the wrong thing. Ship the write-back.


FAQ
What is a company data API?
A search or lookup that returns a structured business record: name, domain, industry, size, location, and a stable id. Optional blocks cover history, funding, or technologies. A miss is a normal result.
What is the best company data API?
The one that resolves a domain you already know, misses cleanly when it should, and writes back to the tool you already run. A homepage that cannot be tried is not in the running.
How do I get company data into a CRM?
Call enrich with the domain or company id you store, then map fields onto the account. Do not overwrite human-confirmed values with an older confirmed_at.
Does the company record include emails?
It should not include a personal inbox. Role addresses, if returned, are unverified until the verification API says otherwise.
How do credits work?
However the contract you sign defines them. Ask whether a miss spends a unit before you compare headline prices. This page does not publish a Prospect Desk credit price.