B2B data accuracy: 3,000 people checked against LinkedIn
How many people in a B2B dataset still work where it says, with the title it says? We drew 3,000 people at random from Datacircle's US dataset (the snapshot of 2026-10-04, 43.6M people) and looked each one up on LinkedIn on 2026-10-08.
88% have the same company and the same title on LinkedIn (95% CI: 86% to 89%), and 93% the same current company. Every number below is a count, and they are all on GitHub as CSV.
The result
| Against LinkedIn | Share | 95% CI | Count |
|---|---|---|---|
| Same company and same title | 88% | 86% to 89% | 2,628 of 3,000 |
| Same current company | 93% | 92% to 93% | 2,776 of 3,000 |
| LinkedIn URL still works | 97% | 96% to 97% | 2,901 of 3,000 |
Where the other 12% are
| What LinkedIn shows | Share | Count |
|---|---|---|
| Same company, same title | 88% | 2,628 |
| Same company, different title | 5% | 148 |
| Different company | 1% | 34 |
| We have a company, LinkedIn has none | 3% | 91 |
| LinkedIn URL dead (profile gone or renamed) | 3% | 99 |
Of the 148 different titles, 36 differ only in wording (one holds the other, or punctuation), 29 are the company's name in our title, and 83 are another title.
By file, and by how old the profile is
| Group | Same company and title | 95% CI | Count |
|---|---|---|---|
| In the 10M+ file | 88% | 85% to 90% | 585 of 665 |
| In the 50M+ file only | 87% | 86% to 89% | 2,043 of 2,335 |
| Updated 0 to 30 days before the snapshot | 75% | 71% to 79% | 299 of 399 |
| Updated 30 to 90 days before | 87% | 84% to 90% | 512 of 590 |
| Updated 90 to 180 days before | 92% | 91% to 94% | 1,761 of 1,906 |
| Updated 180+ days before | 53% | 44% to 63% | 56 of 105 |
A profile last updated 180 days or more before the snapshot has the same company and title 53% of the time, against 92% at 90 to 180 days. 105 of the 3,000 people drawn are that old: their profiles were updated 104 days before the snapshot on average (median 123). The 10M+ file and the rest of the 50M+ file match alike.
Field by field, on the 2,901 profiles LinkedIn still has
| Field | Same as LinkedIn | 95% CI | Count |
|---|---|---|---|
| First name | 97% | 96% to 98% | 2,816 of 2,901 |
| Last name | 95% | 94% to 96% | 2,757 of 2,901 |
| Headline | 92% | 91% to 93% | 2,662 of 2,901 |
| Still in the US | 100% | 100% to 100% | 2,901 of 2,901 |
| State, where LinkedIn names one | 100% | 100% to 100% | 2,532 of 2,536 |
| Current job's start month, same company | 94% | 93% to 95% | 1,455 of 1,552 |
Emails: not judged
27% of the people drawn have an email, verified 26 days before the snapshot on average. LinkedIn doesn't show emails, so this benchmark doesn't judge them.
How it was taken
- Sample: 3,000 people drawn at random from the 50M+ file's people; 665 of them are also in the 10M+ file.
- Reference: each person's LinkedIn profile, asked by LinkedIn URL on 2026-10-08 through a LinkedIn profile API that fetches the profile from LinkedIn again when its copy is stale. Its answers were 2 days old on average when asked (median 0); 91% had been fetched from LinkedIn within the week.
- Same company: our current company is LinkedIn's current company.
- Same title: the two titles are equal once normalized (case, punctuation, accents, the spacing around & and /).
- 95% CI: the 95% confidence interval of each share.
- No person is named here: only counts.
The same benchmark, with every number as CSV, under CC BY 4.0: github.com/waynehamadi/datacircle.
The data, and a profile as it is today
The 10M+ file is the free US B2B leads dataset, every column with its fill rate. To read a person's profile as it is today rather than as the file has it: Right now we have 2 live LinkedIn profile APIs that we trust: Up2Data and HarvestAPI. Live. Each request goes to the provider and gets the profile as it is today. See the LinkedIn profile API.
Questions
How accurate is B2B contact data?
In our October 2026 benchmark, 88% of 3,000 people drawn at random from a US B2B dataset had the same company and the same title on LinkedIn (95% CI: 86% to 89%), and 93% the same current company. 3% of their LinkedIn URLs no longer worked.
How fast does B2B data go out of date?
In the same benchmark, people whose profile was updated 90 to 180 days before the snapshot had the same company and title on LinkedIn 92% of the time; those updated 180 days or more before, 53%.
How was it measured?
3,000 people drawn at random from the 50M+ US file, each looked up by LinkedIn URL on 2026-10-08 through a LinkedIn profile API that fetches the profile from LinkedIn again when its copy is stale. A match is the same current company and the same title once both are normalized (case, punctuation, accents).
Can I reuse these numbers?
Yes, under CC BY 4.0, citing datacircle.dev. Every number is on GitHub as CSV: https://github.com/waynehamadi/datacircle.
Sign up at datacircle.dev with your work email: a $5 credit, that's 4,000 LinkedIn profiles at $1.25 per 1,000. Free: 10M+ U.S. B2B leads, as a flat file. Download it at datacircle.dev.
Sign up