TL;DR
Discovery and enrichment are two different products sold under one word. A discovery endpoint takes a filter set and returns a list of people you did not have. An enrichment endpoint takes an identifier you already hold and returns fields attached to it. Buying one when you needed the other is the most common procurement error in this category. Best for: teams writing the requirements document before the trial.
Apollo.io is the only major database vendor that publishes both per-seat pricing and per-endpoint rate limits, and its credit schedule charges 1 credit for demographics or an email and 9 when a mobile number is returned. Best for: teams that want one vendor covering search, enrichment and sequencing with numbers they can audit.
People Data Labs is the cleanest pure-data API in the category: usage-based credits, no seat fee, and billing only on a successful match. Search charges one credit per profile returned, which makes broad queries expensive fast. Best for: engineering teams building a product or a warehouse on top of licensed data.
ZoomInfo publishes no API price at all, but its documentation confirms search consumes no credits and only enrich does. That single design choice changes how you architect against it. Best for: teams already holding a ZoomInfo contract who want to stop paying for exports twice.
Coresignal and LeadMagic sit at opposite ends of the same idea: both bill purely on credits with no seats, but Coresignal charges 10 to 20 credits per person record while LeadMagic prices from $0.007 and leaves people, company and job search unmetered on its upper plans. Best for: cost modelling before you commit to volume.
Contents
- What a B2B data API actually sells you
- How to read a credit schedule before you sign
- 1. Apollo.io
- 2. People Data Labs
- 3. ZoomInfo
- 4. Coresignal
- 5. LeadMagic
- Side by side summary
- Additional tools to consider
- How to run a fair bake-off
- FAQ: B2B Data APIs
What a B2B data API actually sells you
There are four separable things inside every vendor in this category, and the pricing page almost never separates them.
The first is discovery. You pass filters, such as industry, headcount band, geography and job title, and the API returns people or companies you did not previously have. This is the expensive one, because you are asking the vendor to hand over slices of its index.
The second is enrichment. You pass an identifier you already hold, usually an email address, a domain, a full name plus company, or a vendor-specific person ID, and the API returns fields attached to that record. This is cheaper per call and much easier to forecast, because your call volume equals your record count.
The third is verification. You pass a value and the API tells you whether it is currently deliverable or reachable. Several vendors in this list fold verification into enrichment silently, which means you are paying for it whether or not you use it.
The fourth is licensing terms, and it is the one that decides whether the integration you are planning is even permitted. Some vendors sell you a right to enrich records inside your CRM but not a right to store the returned data in your own warehouse or expose it in a product you sell. That distinction does not show up in a rate limit table.
Write down which of the four you need before the first demo. A team that needs enrichment on 40,000 existing CRM records has a completely different shortlist from a team that needs to build a 400,000-record total addressable market list from scratch, and both of them have a different shortlist again from a team building a product feature on licensed data.
How to read a credit schedule before you sign
Three questions separate a forecastable bill from a surprise.
Does the vendor charge for misses? People Data Labs documents that credits are consumed per successful match, defined as a request returning at least one profile with a 200 status, and that unsuccessful requests are not billed. LeadMagic states it charges only on valid results. Datagma states no match means no credit used. This matters enormously at low match rates: if your list is niche and your hit rate is 30%, a vendor that bills on attempts costs you more than three times what the headline per-credit price implies.
Does search cost the same as enrichment? This is where the vendors genuinely diverge. ZoomInfo's own enterprise API getting-started guide states that using the Search endpoint does not consume credits and that you are only charged a credit when you Enrich a record. People Data Labs does the opposite: its Search APIs consume one credit per profile returned, so a single request can consume many credits, and its documentation says so explicitly. Coresignal splits the difference, documenting that search itself deducts nothing on its DSL and filter endpoints while collection costs 1 credit for a job or post, 10 for a base or clean company or employee record, and 20 for multi-source.
Are rate limits per key, per user, or per team? Apollo.io's documentation is unusually explicit here: limits are per team rather than per API key or per user, all keys and users in a workspace draw from the same limits, child workspaces share the limits of the parent workspace, limits are per endpoint, and three windows are enforced at once at per minute, per hour and per day. If you were planning to buy throughput by issuing more keys, that plan does not work on Apollo.io. Test it against your own account rather than assuming, because Apollo.io also ships a usage-stats endpoint that returns your actual limits.
1. Apollo.io
Layer: Database plus enrichment plus sequencing, sold per seat with a credit meter
Best for: Teams that want documented rate limits and a published price on the same page
Apollo.io is the default first call in this category, and the reason is transparency rather than data quality. All API requests go to a single base URL, users authenticate with an API key in an x-api-key header, and partners building on behalf of mutual users authenticate through an OAuth 2.0 authorisation flow. The company documents a database of over 240 million contacts and 30 million companies on its enrichment reference pages, and separately markets 210 million contacts and 35 million companies elsewhere, so treat the headline count as a marketing range rather than a fixed figure.
The endpoint surface covers people enrichment for one person, bulk people enrichment for up to 10 people in a single request, organisation enrichment, people search, and the transactional endpoints for accounts, contacts, deals, sequences, tasks, calls and conversations.
The credit behaviour is the part worth internalising. Endpoints that create, update, list or manage records cost nothing. People enrichment costs 1 credit for demographics or an email and 9 credits if a mobile number is returned. That nine-to-one ratio is the single most important number in an Apollo.io budget, because it means a workflow that requests mobile numbers by default costs roughly nine times a workflow that does not. Export credits are also consumed on CSV export, CRM push and the Person API, which means the same record can cost you twice if your pipeline exports what it just enriched.
Published rate limits differ by endpoint type and plan. Enrichment endpoints allow 50 requests a minute on Free and 1,000 on Basic, Organization and Professional, with no hourly or daily limit on paid plans. Other endpoints allow 200 a minute and 2,000 a day on Basic and Organization. Query the usage-stats endpoint for your team's actual numbers rather than trusting a third-party summary, including this one.
Pricing: Free at $0 with 900 credits per seat per year released monthly; Basic $49 per seat per month on annual billing with 30,000 credits released upfront; Professional $79 with 48,000; Organization $119 with a three-seat minimum and 72,000. A 14-day trial covers Basic and Professional. The Unlimited Plan carries a cap of 10,000 credits per account per month for non-paying accounts, or the lesser of amount paid divided by $0.025 and 1,000,000 per year.
Compare contact coverage on a sample from your actual account list. Count correct-person emails and reachable direct numbers separately, and retain unmatched records in the denominator. A global database size or an undocumented benchmark cannot establish accuracy in your territory.
Verdict: Buy Apollo.io when you want one vendor, published prices, documented limits and a credit schedule you can model. Do not buy it as your only phone source, and do not architect a pipeline that requests mobile numbers on every record unless you have priced the nine-credit charge.
2. People Data Labs
Layer: Licensed person and company data, API and bulk feed, no seats
Best for: Engineering teams building a product or a warehouse rather than running a sequencer
People Data Labs is the cleanest architectural fit in this list for anyone who wants data as an input to something they are building. There are no seats. There is no application. There is an API, a set of credits and a licence.
The plan structure is short. Free gives up to 100 records or credits per month with easy signup, no commitment, and no contact data included. Pro starts at $98 per month, begins at as few as 350 credits per month, includes contact data and premium fields, offers monthly and annual subscriptions, adds CSV enrichment, and comes with email support. Enterprise is custom priced with formal data testing, annual contracts, flexible payment terms, additional field bundles, access to all APIs, a designated success team and SSO or SAML. You can sign up free and upgrade to Pro directly from the API dashboard with no sales call, and volumes above 100,000 records require contacting sales.
For People Data Labs, scope person and company endpoints, monthly volume and required fields. A contact enrichment allowance is not automatically an equivalent company-discovery allowance.
For data procurement, compare the actual lookup input, accepted output and usage charge. Retain unmatched and wrong-person records in the result set so that apparent coverage does not hide unusable data.
Pricing and assurance: scope the required API volume and fields with the provider. For procurement evidence, request current assurance documentation covering the service being purchased; do not describe data itself as SOC 2 certified.
Where it falls short: Search at one credit per profile returned punishes exploratory querying, which is exactly what a discovery workflow does. If you do not know your filters precisely, you will pay to learn them. Contact data is excluded from the free tier entirely, so the free plan tests the shape of the API and not the thing you are buying. And this is data, not an application: there is no sequencer, no dialer and no UI for a rep.
Verdict: The right answer when data is an input to a system you control and the wrong answer when you need a tool a rep can open. Model your search-to-enrich ratio before you commit, because that ratio, not the per-credit price, determines the bill.
3. ZoomInfo
Layer: Enterprise B2B intelligence with an enrichment API attached to a platform contract
Best for: Teams already under contract who want to stop paying twice for the same record
ZoomInfo publishes no API pricing, and the pricing page is a form. What it does publish is documentation, and the documentation contains the single most useful architectural fact in this article: search consumes no credits, and you are only charged when you enrich. ZoomInfo's own enterprise API getting-started guide states it plainly. A call to the Enrich endpoint returns all data associated with a specific record found through the Search endpoints, and covers contacts, companies, locations, org charts, technologies, hashtags and IP addresses. There is a separate Company Master Data Enrich endpoint that combines ZoomInfo firmographic data with EverString firmographic data and adds attributes.
That design means the correct pattern against ZoomInfo is to search widely, filter locally, and enrich narrowly. Teams that enrich first and filter second are paying for records they then discard.
Coverage and accuracy claims use different denominators. Ask whether a percentage refers to returned records, attempted matches, verified contact fields or actual connections. Keep vendor claims separate from the results of your own sample.
Verdict: If you already pay for ZoomInfo, use the API and stop exporting. If you do not, do not start here unless US mobile coverage is the specific problem you are solving and you have budget in five figures.
4. Coresignal
Layer: Bulk and API professional data, credit-metered, no seats
Best for: Teams that want published volume tiers and predictable rollover
Coresignal is the vendor in this list that publishes the most complete self-serve ladder for data at scale. Credits are a single universal unit across every endpoint and plan, and each plan includes a monthly or yearly allocation deducted per successful request.
The credit schedule is published: an employee record costs 10 to 20 credits, a company record 10 to 20, a job posting 1 credit, and contact enrichment 20 credits. The tier detail matters, because 1 credit buys a job or a post, 10 buys a base or clean company or employee record, and 20 buys multi-source. Search itself deducts nothing on the Elasticsearch DSL and filter endpoints, which puts Coresignal architecturally on the ZoomInfo side of the search question rather than the People Data Labs side.
Eight published tiers run from a 7-day free trial to a $5,000 per month Elite plan, differentiated by credit count, requests-per-second rate limit, access to Search Preview and the Contact Enrichment API, credit rollover and support model. The published waypoints are a free 7-day trial with 2,000 credits, Mini at $49 per month, Starter at $199, Pro at $499 and Growth at $1,000 per month for 150,000 credits, which works out at $0.007 per credit. Growth includes a 20 requests per second rate limit, a dedicated account manager, multi-source endpoints, an Agentic Search API, Elasticsearch DSL, webhooks, search preview, the Contact Enrichment API and a 3-month credit rollover. Annual plans take 10% off and issue the full year of credits upfront. Free trial credits are issued once per company email domain, to the first person who registers.
The AI-search list feature has its own schedule: preview cost is calculated in increments of 20 records at 20 credits per batch, the maximum preview is 100 records so a single preview can cost 100 credits, enrichment is charged per record at 1 to 20 credits, and the maximum enrichment cost is capped at 2,000 credits for a fully enriched 100-record preview list. Searches returning zero results or requiring clarification consume nothing.
Pricing: 7-day free trial with 2,000 credits. Mini $49 per month. Starter $199. Pro $499. Growth $1,000 for 150,000 credits at $0.007 each. Elite $5,000. Annual saves 10% with credits issued upfront. Authentication uses a single apikey HTTP header.
Where it falls short: Ten to twenty credits per person record is expensive relative to LeadMagic or People Data Labs on a straight per-record basis, and the multi-source tier that most buyers actually want is the 20-credit one. The credit cliff between tiers is real: the gap from Pro to Growth is a $501 monthly step. And the free trial is one per domain, which makes a genuine multi-team evaluation awkward.
Verdict: The strongest published ladder in the category for teams that know their volume. Price the multi-source tier, not the base tier, because that is the one you will use.
5. LeadMagic
Layer: Pay-per-result enrichment and search API, one credit pool, no seats
Best for: Small and mid-sized teams that want low unit cost and unmetered search on upper plans
LeadMagic is the aggressive-pricing entrant, and the structural choice that makes it interesting is that people, company and jobs search are free and unlimited on the Professional and Ultimate plans while contact-detail unlocks stay credit-metered. That is a third position on the search question, and for a discovery-heavy workflow it is the most favourable of the three.
The API documents POST /v3/people/search, POST /v3/companies/search including a company-search alias, and POST /v3/jobs/search with vector, facets and deep modes, all marked free and unlimited on the upper plans with the caveat that they are browse only. The published index sizes are 400 million or more people, 25 million or more companies and 46 million or more jobs. There is an OpenAPI 3.1 spec with smoke tests, a hosted MCP server with OAuth, official Claude Code and Cursor plugins, agent skills and credit-safe hooks.
Pricing: Basic $49.99 per month or $490 per year, 2,000 credits, $0.025 per credit. Credits from $0.007 at volume. People, company and jobs search included from Essential and unmetered on Professional and Ultimate. Full API on every plan. Manual and automatic top-ups available.
Where it falls short: Transparency drops at the top of the ladder, where higher volumes move to custom pricing, and the published plan names run ahead of published prices for Essential, Professional and Ultimate. Per-credit pricing is a poor fit for very low volume, where a subscription floor of $49.99 buys more than a small team needs. And the coverage claims are first-party with no independent benchmark published, so a match-rate test on your own list is mandatory rather than optional.
Verdict: The best unit economics in this list for teams under enterprise scale, and the only vendor here that gives away discovery and charges for the reveal. Test match rate on your real list before you commit to volume, because low unit price on a low hit rate is not a saving.
Side by side summary
| Tool | Layer | Best for | Entry price |
|---|---|---|---|
| Apollo.io | Database, enrichment and sequencing, per seat plus credits | Published limits and one-vendor coverage | $49 per seat per month annual |
| People Data Labs | Licensed person and company data, credits only | Building a product or a warehouse | From $98 per month |
| ZoomInfo | Enterprise intelligence, API on a platform contract | Existing contract holders, US phone data | Quote only, three-seat minimum |
| Coresignal | Bulk and API professional data, credits only | Known volume and published tiers | $49 per month, Mini |
| LeadMagic | Pay-per-result enrichment, credits only | Low unit cost with unmetered search | $49.99 per month, Basic |
Additional tools to consider
- BlitzAPI sells flat-rate unlimited access with every endpoint unmetered once subscribed, at $399 per month for Unlimited Leads and $499 for Unlimited Email, against 380 million or more contacts and 60 million or more companies at 5 requests per second.
Field reports on this cohort are mixed and consistent in their mixedness. Operators holding all three of Blitz, QuickEnrich and GetLeads describe most of the data as outdated but priced accordingly, with one summarising that cheap data will never be 100% accurate and that this is fine if you plan for it. A more specific pattern reported repeatedly: QuickEnrich stronger on email coverage, Blitz and GetLeads stronger on lead coverage if you already run a work-email waterfall, and accuracy estimates around 70% to 80% valid with thin coverage of senior titles. Treat the flat-rate tier as a volume layer under a verified layer, not as a replacement for one.
How to run a fair bake-off
The test that decides this category takes a day and almost nobody runs it.
Take 500 records from your actual target segment, not a sample the vendor supplies. Strip them to the identifiers each API accepts. Run the same 500 through every shortlisted vendor on the same day. Then measure four things: match rate, field completeness on the fields you actually use, cost per matched record including misses, and the wall-clock time to process the batch at the rate limit you were sold.
The fourth measure is the one people forget and the one that kills projects. A vendor with the best match rate and a 200-rows-per-20-minutes practical throughput cannot support a daily 5,000-record refresh regardless of how good the data is.
FAQ: B2B Data APIs
What is the difference between a data API and a database subscription?
A database subscription sells a seat in an application with an export allowance. A data API sells calls against an index, usually metered in credits, with a licence describing where the returned data may live. The practical difference shows up in three places: whether you can store what you retrieve, whether you can expose it in a product you sell, and whether your bill scales with people or with volume. Apollo.io sells both models at once, which is why its credit schedule charges export credits on CSV export, CRM push and the Person API alike.
Do these APIs charge for lookups that return nothing?
Several explicitly do not. People Data Labs documents that credits are consumed only on a successful match returning at least one profile with a 200 status and that unsuccessful requests are not billed. LeadMagic states it charges only on valid results. Datagma states no match means no credit used. Coresignal states searches returning zero results or requiring clarification consume no credits. Where a vendor does not say, assume it bills on attempts and price your expected miss rate into the model, because on a niche segment the miss rate is the dominant cost.
Which is cheaper, discovery or enrichment?
Enrichment, almost always, and by a wide margin. The reason is that enrichment volume equals your record count while discovery volume equals your curiosity. The three vendors here take three different positions: ZoomInfo documents that search consumes no credits and only enrich does, People Data Labs charges one credit per profile returned by search, and LeadMagic leaves people, company and jobs search unmetered on Professional and Ultimate while metering the contact-detail unlock. Match the vendor to the shape of your workflow rather than to the per-credit price.
How do rate limits actually constrain a real pipeline?
Read the enforcement model, not the number. Apollo.io documents that limits are per team rather than per key or per user, that child workspaces share the parent workspace's limits, that limits are per endpoint so people search and contact creation do not share counters, and that per-minute, per-hour and per-day windows are enforced simultaneously. Coresignal publishes requests per second by tier, reaching 20 per second on Growth. Neither of those is knowable from a pricing page, and both decide whether a nightly refresh finishes before morning.
Should we buy one API or waterfall several?
Add a provider when it fills a recurring, valuable gap. Measure incremental correct matches after deduplication and stop steps that add cost without useful coverage.
What licence terms should we check before building on one of these?
Four. Whether you may store returned data in your own warehouse rather than only in a connected CRM. Whether you may expose it in a product you sell to third parties. Whether the licence survives contract termination for data already retrieved. And what the deletion obligation is when a subject exercises a data right. People Data Labs is explicit that Enterprise covers API or data licence feeds with custom contract and usage terms, which is the tier where these questions get answered. On self-serve tiers, the answers are in the terms of service and they are frequently narrower than buyers assume.
The source is one part of the data workflow. Our guide to b2b data enrichment tools covers the adjacent options.
Work with Forma Nôrden
We build signal based outbound systems for B2B companies selling into the enterprise and upper mid market. Choosing between a discovery API and an enrichment API is a data architecture decision, not a procurement one, and getting it wrong shows up as a bill that scales faster than the pipeline. Explore how we work.
For enquiries about this article: partnerships@formanorden.com
