Decision guide

Choose a company enrichment API for scoring leads from a domain

Five company enrichment routes called on the same ten domains: which lead-scoring fields each one actually returned, what each record cost, and how far their employee counts disagree.

Written for people asking: What are the best company enrichment APIs for scoring inbound leads automatically?Choose a company enrichment API from a company domainWhich company enrichment API returns the fields a lead-scoring rule needs?

Decide on fields before price

An inbound scoring step reads a handful of fields and turns them into a route: size bucket, country, industry, sometimes revenue, funding or tech stack. A company record that comes back as a hit but with the size field empty or wrong scores the lead wrong, and nothing in the bill tells you. So the questions, in order:

1. Which fields does the rule read? A rule on size and country needs far less than one that reads funding stage or a technology in use.
2. Does the provider return those fields filled, for companies like yours? “80+ datapoints” on a pricing page is the schema, not the fill rate.
3. Is the headcount a number or a band, and does it agree with anyone else? Bands are coarser than they look, and two providers here returned exact counts outside their own band.
4. Then price and billing shape. Every route we called bills per company returned; the no-subscription guide covers what that costs at irregular volume.

The providers and their terms

Terms as published on each vendor’s pricing and docs pages on the date shown. “Per company enriched” is the entry paid tier divided by its credits, times the credits one company costs; it is list-price arithmetic, not an observed cost.

What each route returned for the same ten domains

On 17 September 2026 we sent the ten company domains from our fixture (Stripe, Vercel, Shopify, Canva, Figma, Linear, Notion, Replit, Intercom, Front) through Treg’s routed company endpoint five ways: once unpinned, and once pinned to each of Hunter, Apollo, Findymail and People Data Labs with the waterfall off, so each answer is that provider’s own record. People Data Labs was run on the first five domains only, because at $0.38 a record ten would have used most of the run’s budget. A count is how many records had the field filled; a dash means the field was not in any response we received.

Every call was a hit, so the per-company figure is the total billed divided by the records returned. Hunter’s is under its $0.0049 rate because Stripe had been looked up minutes earlier and was billed a tenth; Hunter’s own terms say repeated searches in the same month are not re-charged.

Three things the counts hide:

• Hunter’s NAICS code is coarse. Nine of ten records carry the sector code 51 and the same six-digit code, 519130, for companies as different as Stripe and Linear. It fills the column without separating the leads.
• The Companies API’s records are old. Its sync stamps on these ten ran from August 2024 to October 2025. Hunter’s were all indexed 17 May 2026. People Data Labs names a dataset version (35.2), not a date.
• Apollo’s record is the heaviest by far. Stripe alone came back with 426 technologies, 17 funding events and headcount by department. That is useful for scoring and a lot to pass into an agent’s context; select the fields before the model sees them.

Contact details on the company record

A company record carries the company’s own contact fields, not a named person’s email. Counted from the same receipts:

• Apollo: a main phone number on 8 of 10, a street address on 9, a Twitter/X link on 9.
• Hunter: a phone number on 6 of 10, and a list of 5 to 100 email addresses it had found on each domain, on all 10. They are masked in the published receipts, so we cannot say how many were role inboxes.
• The routed record (The Companies API): a phone number on 1 of 10, an email pattern with its usage share on 4, a Twitter/X link on 8.
• People Data Labs: Twitter/X and Facebook links on 5 of 5, no phone.
• Findymail: no phone, email or social fields.

To reach a named person, use an email finder on a name and domain (contact enrichment guide).

How far the headcounts disagree

The field most scoring rules lean on is the one the providers agree on least. Exact counts where the provider returned one, otherwise its band:

Put into four scoring buckets (1-200, 201-1,000, 1,001-5,000, 5,001+), every provider that answered agreed on only 4 of the 10 domains: Stripe, Shopify, Replit and Intercom. The outliers are not one provider’s fault. Findymail put Figma at 11-50 and Notion at 501-1000; The Companies API put Linear and Front at 1-10; Hunter put Canva at 18,214 where Apollo says 5,500. On five records the provider’s exact count fell outside its own band (People Data Labs for Stripe, Vercel and Canva; The Companies API for Stripe and Intercom), so a rule reading the band and a rule reading the count get different answers from the same response.

The other fields disagree less, but not never. For Notion, Hunter and Apollo said United States, The Companies API said Canada and Findymail said Israel. For Figma, three providers said founded 2012 and The Companies API said 1999.

Who each option fits

• A rule on size, country and industry, run on every signup: Findymail if you already pay for its email finder, Hunter if you want the cheapest record with an exact count. Cross-check the size field on a sample of your own leads first; both had a large miss in ten.
• A rule that also reads revenue, funding, tech stack or growth: Apollo was the only route here that filled all four on the same record. Through Treg it was $0.026 per company with no plan; on Apollo’s own Basic plan the arithmetic is $0.0196, with a per-user annual subscription.
• Funding history and headcount by country: People Data Labs returned both on every record. Through Treg that was $0.38 per company, pay per hit; People Data Labs’ own Company Pro plan works out at $0.10 per record if you use its $100 monthly allowance.
• The cheapest possible record, to decide whether a lead is worth a second call: Treg’s routed endpoint answered for $0.0019 through The Companies API. Use it as a filter, not as the scoring source: it put Linear and Front, which the other providers counted at 180 to 570 employees, in its 1-10 band.
• Two providers, cheaply: Hunter plus Findymail for the same domain is about $0.025 at their rates in this run, and where they disagree on the size bucket the lead is worth a human look.
• Prospeo, Icypeas, Lusha, Explorium, Enrichlayer, Crustdata: not measured here. Every Prospeo-pinned call hit a rate limit on 17 September (below); Treg’s route to Icypeas needs a company LinkedIn URL rather than a domain; the rest were not called. Their terms are in the table above with links to their docs.

What this page cannot tell you

• Which headcount is right. There is no ground truth in this run. The table shows disagreement, and disagreement is enough to say a single source is risky for size-based routing; it is not enough to say which provider is accurate.
• Fill rates on your leads. Ten well-known, venture-funded software companies are the easy case. Fill rates for a 30-person agency or a company outside North America will be lower on every provider, and by different amounts.
• Anything about the providers not run. The empty rows in the fields table are not a verdict on them.
• How the routed provider will behave next week. Treg’s route tries the cheapest eligible provider first. On 14 September that was Hunter; on 17 September it was The Companies API. The response names the provider that served, so a scoring step can log it, but the field shape changes with it.

One thing the run did show about routing: the Prospeo-pinned calls got HTTP 429 (rate limit) on all ten, and with the waterfall header off they still fell through to The Companies API, because a rate-limit error is not treated as a miss. If a scoring step must use one provider’s record, check _treg.served_by in the response rather than trusting the preference header.

Next step

• Run the same comparison on twenty of your own recent inbound domains before you write the scoring rule. The runner and scorer scripts take any list of domains and any set of providers; this run cost $2.42 in total, most of it the five People Data Labs records.
• The route headers used here are documented at treg.to/docs.
• If volume is small and irregular, the no-subscription guide works through what 100 companies a month costs on each billing shape. If you also need a contact at the company, the contact enrichment guide covers finders.

Relationship: the people who run this site also work on Treg, one of the routes compared on this page. Every Treg figure here is a call we paid for, with the raw response downloadable, and the same fixture is run against the alternatives where their terms allow.

Spotted an error or a changed price? Email corrections@enrichmentapiguide.com with the page URL and a source. Corrections are logged on the corrections page.