A B2B database is a structured, searchable store of companies and the people who work at them, with contact details, firmographics and sometimes intent signals, sold either as a subscription with filters and exports or as a one-off file. The subscription vendors report 90 to 95% email accuracy; independent verification of their exports finds 10 to 22% invalid, because records age between the vendor's check and yours.
A subscription database is a product you log into: filters, saved searches, enrichment, exports metered by credits. ZoomInfo, Apollo, Cognism, Lusha and Kaspr are the names that come up. You pay for access and for exports, and the vendor keeps crawling.
A "B2B database for sale" in the other sense is a file: a CSV of some thousands or millions of records, priced per record or as a flat fee, delivered once. Brokers, compilers and marketplaces sell these. The data may originally have come from the same places; the difference is that nobody refreshes it after delivery.
The evaluation differs. For a subscription, you test the export accuracy and the filter coverage for your segment. For a file, you test a random sample and the age per record. Both tests end in the same place: verification of what you actually receive.
Claimed accuracy is vendor-reported. Verified figures are measured only for Apollo and ZoomInfo (our export studies); the other rows are estimates from the finder accuracy range of 10 to 25% invalid.
| Database | Records (vendor-reported) | Claimed email accuracy | Invalid at export (measured or estimated) | Strength | Weak spot |
|---|---|---|---|---|---|
| ZoomInfo | 260M+ professionals | 95% | 12 to 20% | US enterprise coverage, org charts, intent | Price; mid-market and EU gaps |
| Apollo | 275M+ contacts | 91% | 5.8% invalid + 18.4% catch-all | Price, built-in sequencing, find rate | Database age; "verified" badge is historical |
| Cognism | 400M+ profiles | 98% on Diamond-verified phones | 10 to 18% | EU and UK coverage, GDPR tooling, phones | Email accuracy below phone accuracy |
| Lusha | 100M+ contacts | 81% (self-reported) | 15 to 22% | Chrome extension speed, SMB price | Thin outside US and Western Europe |
| Kaspr | 500M+ profiles (via LinkedIn) | Not published | 12 to 20% | LinkedIn workflow, France and DACH | Phone-first; email coverage thinner |
Export studies: Apollo, 100,000 contacts, ZoomInfo accuracy. Reviews: Cognism, Lusha. Head to head: Apollo vs ZoomInfo, Apollo vs Cognism.
Three hundred million records is meaningless if your segment is 4,000 logistics managers in the Benelux. Run your exact filter on a trial and count. Then export 200 and verify.
Vendors re-crawl on a schedule. Ask how old the median record in your segment is, and whether a "verified" flag carries a date. Apollo and ZoomInfo both show one if you look.
Seat price plus credits per export. Work out cost per confirmed valid address: list price divided by (exports times share valid after verification). A cheaper seat with 22% invalid can cost more than a dearer one at 12%.
EU and UK selling needs a vendor that documents lawful basis per record and handles opt-outs upstream. Cognism and Kaspr lead here; US-first vendors vary.
Database exports are overwhelmingly business domains. This is what business-domain addresses verify at in the BounceZero corpus, April to October 2026, compared with freemail addresses that leak into "business" exports.
| Address type | Verified | Invalid | Catch-all | Unknown | Confirmed valid |
|---|---|---|---|---|---|
| Non-freemail (business and ISP) domains | 6,514,884 | 14.5% | 2.3% | 27.0% | 32.6% |
| Freemail | 1,415,154 | 17.6% | 1.1% | 2.1% | 74.0% |
| .com business | 2,158,955 | 16.4% | 4.4% | 14.7% | 59.7% |
| .org | 56,401 | 24.6% | 12.4% | 25.0% | 36.2% |
| .edu | 17,961 | 32.8% | 8.3% | 18.5% | 38.2% |
The 27% unknown on non-freemail (business and ISP) domains is the number no database reports. It is corporate gateways and ISPs refusing to confirm a mailbox, which is why a verifier that labels unknown honestly is worth more than one that guesses.
When the "database" is a file delivered once, the subscription questions do not apply. These do.
A single "last updated" date for the whole file is a red flag. Compilers merge sources of different ages; you need the distribution.
Ask for a unique-person count. Compiled files commonly carry 5 to 15% duplicates across sources.
Two hundred rows chosen by a random seed you supply. Under 8% invalid and under 10% catch-all plus unknown is a usable file.
Title, phone, company size and industry fill rates by row. A database with 40% missing titles cannot be segmented and should be priced as a raw list.
Observed addresses versus pattern-inferred, and the original sources. If the vendor will not say, the file is probably scraped or recycled.
Hard-bounce replacement within 30 days; licence for single or multiple uses; whether you may re-verify and keep the clean subset indefinitely.
Whatever database you are considering, export or request 200 random rows and verify them before you commit. Every BounceZero account includes 100 verifications a month free, and the Starter pack is 2,000 credits for $6. You get valid, invalid, catch-all and unknown per row, with unknowns refunded. Compare the result to the thresholds above and to the vendor's claim. The gap is your negotiating position.
Verify every export on the day it is pulled, not when the campaign is ready; the two can be weeks apart. Route valid rows to sequences, catch-all rows to a warm low-volume track on a secondary domain, unknowns to a re-check or a phone list, and invalids back to the vendor for credit.
Re-verify open leads before every send. The list decay study measured mailbox loss month by month; a 90-day cadence keeps a database-sourced list under the 2% bounce line. The contact database guide covers the full workflow from filter to send.
Providers compared on coverage, accuracy and compliance
What a leads database should contain and how to score one
The subscription vendors in depth
The corpus tables behind this page
What to expect, row by row
Sources ranked by data quality
It depends on segment and geography. ZoomInfo for US enterprise with intent data, Apollo for price and built-in sequencing, Cognism for UK and EU coverage with compliance tooling, Lusha for small teams working from a Chrome extension. Test each on your exact filter with a 200-row export and verify it; the vendor with the most confirmed valid addresses in your segment per dollar wins.
Vendors report 90 to 95% email accuracy. Independent verification of exports finds 10 to 22% invalid at export time, because records age between the vendor's check and yours and because catch-all domains are counted as deliverable. Phone accuracy is typically higher than email accuracy at the vendors that verify phones manually.
Subscription seats run from about $50 a month at the low end to over $1,000 a month for enterprise tiers, vendor-reported, plus credits per export. One-off files run from a cent to sixty cents per record. The useful metric is cost per confirmed valid address, which only verification gives you.
Yes, from brokers, compilers and marketplaces, as a file. It arrives once and nobody refreshes it, so evaluate it on age per record, unique-person count, field fill rates and a random sample you verify yourself. Subscription vendors generally do not sell their full database as a file.
Buying is legal; using it depends on lawful basis. For corporate contacts you can rely on legitimate interest under UK GDPR, with a privacy notice naming the source and an opt-out honoured promptly, and PECR permits B2B marketing email to corporate subscribers. The vendor must be able to show how each record was collected. Sole traders are individuals and need consent.
Two hundred rows from any database, valid, invalid, catch-all and unknown back. Compare to the vendor's claim and negotiate from there.
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.
Where lists come from, what bounces, and what to check before you pay
Where to buy leads for business in 2026
Buy leads for real estate in 2026
Business leads lists for sale in 2026
Real estate email address lists in 2026
Buying an email list for lead generation in 2026
A B2B leads database is only as good as the share of its contacts you can still reach
Continue through related topics