Methodology, protocol v1
How we measure B2B data accuracy
Six steps, the same for every provider, ours included. If a number on any report disagrees with this page, this page is the one that is wrong and we will fix it.
In plain words
- We send the same list of people to every B2B data provider we test, 9 in the latest run on 2,000 LinkedIn profiles, each through its own API.
- For each person, we ask one question: does the provider know where they work today, and what their job title is?
- We check every answer against the person’s public profile, read on the day of the test from a source that is none of the providers tested.
- An answer counts as accurate only when both the company and the job title match. Out of every 100 people sent, we count how many come back accurate: that is the number providers are ranked on.
- The rules are the same for every provider, BlitzAPI included, and every result is published.
1. How a test runs, step by step
One run is one photograph: the same list, the same rules. Here is the order in which it happens, and nothing happens outside it.
- 01
The list
A run starts from lists of LinkedIn profile URLs that mix people long settled in their job with people who just changed. Of the 2,000 people in the latest run, 260 had started their current job 3 months ago or less, and 940 more than 12 months ago; every provider named an outdated company more often for the first group than for the second. Every report shows its results by how long ago the current job started.
- 02
Who is left out, before anyone is queried
Contacts are removed before any provider sees the list, never after the answers are in. In the latest run nobody was removed: every profile on the list was in scope. Every report publishes how many contacts were removed under each cause.
- 03
A reference that belongs to nobody
On the day of the test, the current employer and title of every contact are read from their public profile, from a source that is none of the providers tested. Only fields any visitor can see are read, no contact details enter the reference, and LinkedIn is not a party to these tests. A run only proceeds when the reference is essentially complete. Measured on November 13, 2026 on 2,000 profiles, the latest run read 2,000 public profiles first, 100% of those requested, and all 2,000 were then sent to every provider.
- 04
Same list, same day
Every provider receives the identical list, one call per contact, through its own public API, with the person as input, never a name and a company to search. The LinkedIn ids and the profile URL are both tried with every provider, and the run sends each one a single identifier: in the latest run, the LinkedIn member id for BlitzAPI, the LinkedIn public profile id for Hunter and the LinkedIn profile URL for the other seven. Transport errors are retried until the run is clean. A contact a provider genuinely does not have counts against coverage, not against accuracy.
- 05
One verdict per answer
Employers are matched by identifier first, then by URL, then by normalized name, because two companies can share a name and one company can be written five ways. When the employer matches, the title is compared after normalization. Each returned contact ends in exactly one of four verdicts; a contact a provider never returned is in none of them and counts against coverage only.
- 06
Recomputed, then published whole
Before anything goes online, a separate script reads the raw answers again, recomputes every verdict and every aggregate, and checks that they add up. Its result is published per provider, warnings included: when a provider passes with a warning, the report says which check and what it affects.
2. Why the test is built this way
Every step above closes off a way of getting a flattering number. These are the six questions worth asking of any benchmark, including this one.
Why not score against a provider’s own database?
A benchmark scored against one vendor’s database measures similarity to that vendor, not accuracy. The reference has to come from outside all of them, or the result is a measure of agreement.
Why hand every provider the person, instead of a name and a company to search?
Every provider is handed the person: the LinkedIn ids and the profile URL are both tried with each of them, and the run sends each one a single identifier (in the latest run, the LinkedIn member id for BlitzAPI, the LinkedIn public profile id for Hunter and the LinkedIn profile URL for the other seven). The test measures the record behind a known person, not the ability to find that person from a name and a company. Mixing the two would measure two things at once and hide which one gave way.
Why remove contacts before the answers come in?
Excluding contacts after seeing the answers is the easiest way to shape a benchmark, and it is the first thing worth checking in anyone else’s. The filter runs before the queries, and the count removed under each cause is published.
How are prices compared?
Prices are not measured: they are list prices, read and dated on each provider's own pricing page. The comparison pages show the monthly bill to enrich 50,000, 200,000 and 5,000,000 people a month (200,000 by default), each time the cheapest self-serve way to buy it, and the cost of 1,000 accurate records, which divides that bill by the accurate records each provider returned in the benchmark. The report itself publishes no price.
How are small gaps read?
Every page reads the result as the test measured it: the higher value ranks first, however small the gap, and the gap is written in points. Three rates also carry a 95% Wilson confidence interval, because they are proportions of a known base: coverage, accuracy, and accurate records per 100. Those intervals are published with the exact numbers, so anyone can see how close two providers are. A segment under 30 contacts is not published at all, one under 50 is published but not ranked, and a segment whose value is unknown (no industry or company size on record) is not published.
Why publish a test we are in?
We sell one of the products measured. The reference comes from none of the providers, every verdict is recomputed by a separate script before publication, and the segments where our own product falls behind are published in the same tables as everyone else’s. That reduces the conflict; it does not remove it. Measure us on your own list before believing any of it.
3. What each measure means
Seven words carry every measured figure we publish. Each one in plain words first, then precisely. Prices are not measured: how they are compared is in section 5.
- Coverage
How often the provider finds the person at all: the share of the people we sent for which it returned someone.
Contacts returned divided by contacts queried. A returned contact counts here even when its data is wrong.
- Accuracy
When the provider answers, how often it is right: the share of its answers with the right current employer and a matching job title.
Right employer and title divided by contacts returned. Right employer means the employer the person's public profile showed as current on the day of the test.
- Accurate records per 100 contacts
Out of 100 contacts you send, how many come back usable. Coverage multiplied by accuracy.
Right employer and title divided by contacts queried, times 100. This is the ranking measure, because it is the only one a provider cannot improve by answering less often or by answering more loosely.
- Stale employer
The provider names an employer the person has left.
Split into lag, where the named company really appears earlier in the person's history, and never worked there, where it does not appear at all.
- Unverifiable
Neither side carries enough data to judge the answer.
No usable company on the provider side, or no current role in the reference. Published so the four verdicts always add up to the contacts returned.
- Work history recall
How much of a person’s career the provider knows, not just the current job.
Reference experience rows matched in the provider answer, divided by reference experience rows, over the profiles the provider returned. Read it beside the number of roles an endpoint returns at all: an endpoint built to return one role scores low on scope, not on quality.
- Margin of error
How far a measured rate could move if the same test ran on another list of the same size.
Half the width of the 95% Wilson confidence interval of a rate. It is published with the exact numbers; the pages rank on the measured values.
The four verdicts
| Verdict | What it means | Base |
|---|---|---|
| Right employer and title | The current employer matches the reference, and the title on that role matches after normalization. | Contacts returned |
| Right employer, different title | The current employer matches, the title does not, or is missing on one side. | Contacts returned |
| Stale employer | The current employer does not match the reference. Usually a real previous role, sometimes a company absent from the history entirely. | Contacts returned |
| Unverifiable | No usable company data on one side, or no current role in the reference. | Contacts returned |
A contact a provider never returned is not in these four: it counts against coverage only. The four verdicts always add up to the contacts returned, which is why the shares on a report are shares of returned contacts and say so.
4. What this test does not tell you
The reference is one day old, by construction
A provider that indexes a change the day after the test is measured as wrong, and it is. That is why accuracy is published by job age rather than as a single number, and why a run is repeated rather than cited forever.
The population is chosen, and it is not the market
The lists are drawn to include the situations that break contact data, such as recent job changes. A benchmark built on long settled contacts only would produce higher numbers for everyone.
The test measures the record, not the search
A provider whose strength is finding someone from a name and a company is not measured on that strength here. Every provider is handed the person.
Providers are not always queried on the same day
When a provider is queried after the day of the reference, the run conditions of the report show it for that provider, and the extra time works in its favor, not against it.
A title match tolerates wording
Titles are compared after normalization, so a rewording counts as a match and a real promotion does not. The direction of title mismatches is measured and published.
5. How prices and limits are compared
Accuracy is measured. Prices, rate limits and access are not: they are read on each provider's own pricing page and documentation, and every page that shows them links to its sources. Prices and limits read on November 15, 2026.
- List prices: in US dollars, before tax, on monthly or annual billing, as published. A provider that prices per seat is priced per seat.
- What one lookup costs: the credits or records a person enrichment uses, as the provider documents it, through the cheapest route that returns the current employer and job title. When a rule is not published, such as whether a lookup that finds nothing is charged, we take the reading that makes that provider cheaper for the buyer.
- The monthly bill: the cheapest self-serve way to enrich a given number of people a month (50,000, 200,000 and 5,000,000; 200,000 by default): one plan, up to 50 seats, or add-on credits at their published price. A pace or a daily cap that cannot deliver the month's volume rules a plan out. With no self-serve option left, the provider shows as a custom plan.
- The cost of 1,000 accurate records: the monthly bill divided by the accurate records it buys, from each provider's accurate records per 100 contacts in the latest edition. It is the only figure that combines a price with a measure.
- What it leaves out: discounts negotiated on a call, add-on packs with no published price, taxes and promotions. Plans named with a word we do not use on this site are called by their price or by the name in the provider's documentation.
Prices change. Each provider's file is read again for every edition, and the site refuses to build when a file is more than 45 days older than the edition it sits next to. A price we got wrong is a correction, under the rules below.
6. What we publish, and how to report an error
Every measured figure is published on the report page itself, and the same report is downloadable as a PDF, free and without a form. The PDF is the page, printed, and every figure in it comes from the same data. No row in either identifies a person. Prices are not measured, so they are not in the report: they appear on the comparison pages, read and dated on each provider's own pages.
Row level files are a different matter. Each row describes a real professional profile and carries a judgement about their current job, so they are not published. The full rule set behind a run is sent to any provider in it, on request, with the test id.
A person who appears in a reference can ask to be removed from future runs by writing to antoine@blitz-api.ai.
Corrections and replies
- Corrected: a wrong value, a calculation error, a billing rule we applied wrongly, a rule described here that the code does not implement. Corrections are versioned and stay visible.
- Published as a reply rather than a correction: disagreement about whether the measure is the right one, or about the population. The objection goes next to the number it contests.
- What we will not do: quietly edit a published figure, remove a list after seeing the results, or delay a run because it came out badly for us.
Write to antoine@blitz-api.ai with the test id of the run. We acknowledge within two business days and correct or retract within five.
Every edition, and the protocol it ran on
A new report every quarter, when the run separates providers and the protocol has not moved mid-flight. The version number moves only when a change would alter a figure already published, if the earlier run were scored again.