A B2B business data service sells or licenses company and contact records: firmographics, org charts, work emails, phones and intent signals, delivered through a platform, an API or a one-off list. Providers differ on where the data comes from, how often it is refreshed and how they handle compliance. What they share is decay: records go stale at roughly a decay rate at which only 19.2% of addresses remain valid after 90 days (BounceZero decay study) for contacts, so the accuracy of any service is a function of its refresh cadence and of whether you verify at the moment of use.
Every provider publishes an accuracy number, and the numbers are not comparable. One measures the share of records with any email; another measures the share of emails that passed its own check when indexed; a third measures phone connect rates from a research team's calls. None measures what you will get: the share of exported emails that accept mail on the day you send.
Three questions make the claims comparable. When was the record last touched, not the database as a whole? What happens to a record the provider cannot confirm: kept with a lower score, or dropped? And is the email check an SMTP probe at indexing, or a syntax and MX pass? The answers predict bounce rates better than the headline figure.
The fourth question is the one this page can answer with data: what does a list from this kind of source look like after independent verification. The BounceZero corpus, April to October 2026, covers 7.93 million addresses from customers' lists, and the cuts by domain type and country are in the tables below.
Email accuracy figures are the ranges BounceZero has published from verified exports; vendor-reported means the vendor's own claim. Refresh and coverage are vendor-reported throughout.
| Provider | Data source | Refresh claim | Email accuracy | Compliance posture | Commercial shape |
|---|---|---|---|---|---|
| ZoomInfo | Contributed network, research team, web | Continuous (vendor-reported) | Vendor-reported 90%-plus | GDPR and CCPA programmes; opt-out portal | Annual contract, five figures |
| Cognism | Research team, partner data, phone verification | Diamond Data mobiles re-verified by humans (vendor-reported) | Vendor-reported; mobiles are the headline | EU-first; DNC screening in several countries | Annual contract |
| Apollo | Contributed network plus pattern inference | Continuous | 70 to 82% | Opt-out portal; US-centric | Free tier, per seat from about \$50 |
| Dun and Bradstreet (Hoovers) | Registry data, D-U-N-S numbers, trade data | Firmographics strong; contacts secondary | Vendor-reported | Long-established compliance programme | Annual contract |
| HubSpot Breeze Intelligence (Clearbit) | Web crawl, partner data | Continuous for firmographics | Vendor-reported | HubSpot DPA | Credits inside HubSpot |
| Lusha | Community-contributed, extension | Continuous (vendor-reported) | 70 to 80% | ISO 27701 (vendor-reported); opt-out | Credits, free tier |
| Kaspr | Database plus pattern, EU focus | Continuous | 75 to 85% | GDPR-aligned (Cognism group) | Credits |
| UpLead | Aggregated plus real-time check at export | Real-time email check on download (vendor-reported 95%) | Vendor-reported; the check is at export, which is the right moment | Opt-out | Credits, monthly |
| Seamless.AI | Real-time search across web sources | At search time | Vendor-reported | Opt-out | Per seat, annual |
| RocketReach | Database | Periodic | 65 to 78% | Opt-out | Monthly plans |
| Custom list builders (agencies, Upwork) | Manual research, scraping, finders | At build time only | Varies widely; verify the sample | Your responsibility | Per record or per project |
UpLead's model of checking at download time is closer to verification than the others; it still stops short of labelling catch-all and unknown separately, which is the information you need for routing.
Corpus cuts by domain type and country for the lists data services produce. Source: BounceZero corpus, April to October 2026; full tables in the lead list quality benchmarks.
| Cut | Addresses | Invalid | Catch-all | Unknown | Confirmed valid |
|---|---|---|---|---|---|
| Non-freemail (business and ISP) domains | 6,514,884 | 14.5% | 2.3% | 27.0% | 32.6% |
| Freemail on B2B lists | 1,415,154 | 17.6% | 1.1% | 2.1% | 74.0% |
| .com | 2,158,955 | 16.4% | 4.4% | 14.7% | 59.7% |
| .uk | 21,194 | 17.0% | 3.7% | 10.6% | 65.0% |
| .de | 2,095,440 | 13.9% | 0.0% | 41.2% | 16.5% |
| .fr | 119,582 | 12.1% | 0.5% | 9.3% | 47.6% |
| .it | 1,193,715 | 1.3% | 0.0% | 3.2% | 64.2% |
| .nl | 87,309 | 21.9% | 0.2% | 4.5% | 54.6% |
| .au | 560,378 | 36.5% | 6.3% | 17.6% | 37.3% |
| .ca | 7,561 | 15.9% | 6.2% | 13.9% | 54.4% |
The unknown column is a property of the receiving infrastructure, not of the data vendor: German and some US ISP gateways do not answer probes. A provider that reports no unknowns on a German list is guessing.
Strong on org charts, firmographics, intent and compliance paperwork. Weak on the moment-of-use problem: the contract is annual, the record was touched at some point in the year. Verify exports before sequences; the contract does not do it for you.
Strong on price and speed; a rep can enrich from a LinkedIn profile in a click. Weak on consistency across segments: accuracy varies from the high 70s on large US companies to the low 60s on small firms and non-English markets.
Strong on find rate for obscure contacts, because it searches rather than looks up. Weak on confirmation: a search result is a guess with a score. Treat the output as a finder export and verify it.
Strong on fit: the list is built to your spec. Weak on everything else unless you specify the verification step in the brief and check a 200-row sample yourself before paying the balance.
Accuracy is segment-specific. A sample of US enterprise VPs tells you nothing about French SMB owners. Ask for 200 to 500 random rows from your actual filter.
Run it through BounceZero. Under 10% invalid and a clearly labelled unknown share is a good sample. Over 20% invalid, or a sample with no unknowns on a market where unknowns are expected, is a reason to walk away.
Price per record is the vendor's number. Price per confirmed deliverable contact is yours. A cheaper list at 30% invalid costs more per usable record than a dearer one at 8%.
Verify exports at the moment of use, through the API on CRM import and on every batch before a sequence. Re-verify anything older than 90 days.
For UK and EU contacts, record the legitimate-interest basis, the provider's source statement and your opt-out handling. A good provider gives you the paperwork; it does not take on your responsibility.
How databases are built and what to ask
What the enterprise database returns
US self-serve versus EU research-led
List sellers compared with sample-check rules
What to expect from a phone-first provider
Sources ranked by data quality
No independent benchmark covers all of them on the same segment. Vendor claims run from 90% upward; verified exports from the self-serve platforms land at 70 to 85% deliverable. The honest answer is to verify a sample from your own segment: accuracy varies more by segment and country than by vendor.
A data service is a platform or API you query on an ongoing basis, with refresh and compliance programmes behind it. A lead list is a one-off export, usually from a reseller, that stops being refreshed the moment you buy it. The service costs more and decays slower; the list is cheaper and needs verification immediately and before every send.
Yes, at the moment of use. Enterprise databases verify at indexing time and refresh on their own cadence; B2B contacts decay fast, with only 19.2% of addresses still valid 90 days after verification in the BounceZero decay study between touches. A verification pass on each export costs a few dollars per thousand and keeps bounces under the 2% threshold mailbox providers enforce.
Ask for 200 to 500 random rows from your exact filter, not a showcase set. Verify them. Look at three numbers: invalid share (under 10% is good), catch-all share (route separately), and whether unknowns appear at all on markets where they should (Germany, Australia). A sample with zero unknowns on a German list has been guessed clean.
Selling and using business contact data can rest on legitimate interest under UK GDPR and the EU regulation when the purpose is relevant to the person's role, they are told where the data came from, and objections are honoured. Providers such as Cognism and ZoomInfo publish their basis and run opt-out portals. The responsibility for your use of the data stays with you.
Upload the provider's sample, get valid, invalid, catch-all and unknown back, and negotiate on the number that matters. 100 free checks a month, unknowns refunded.
Ayoub built BounceZero's 5-stage validation pipeline, its dedicated BGP-announced IP infrastructure, and the Patroni HA PostgreSQL cluster behind every verification. Previously built high-volume email delivery infrastructure. Trained at 1337 Benguerir (École 42 network, 2019). Open-source: bgp_analyzer.
Stacks by stage and vertical, with data accuracy from real bounce rates
Prospecting tools for sales outreach in 2026 organised by stage
Data enrichment tools for 2026 compared by what they append (firmographics, technographics, contacts), how...
SaaS lead generation tools for 2026 by motion
Sales leads list software for 2026
B2B email marketing tools for 2026
AI lead generation tools in 2026 sorted by what the AI does
Continue through related topics