A contact number database is a searchable store of business phone numbers linked to named people and companies, sold as seat-based search access or export credits. Coverage is easy to claim and hard to check: email accuracy at export ran 10 to 25% invalid across the measured provider studies, and phone numbers decay faster than email when people change role or move.
A contact number database sells three different things under one label: company switchboards, direct dials and mobile numbers. They behave differently, cost differently and carry different legal risk, so a single coverage number hides the only comparison that matters.
Switchboards are tied to the company record and stay accurate until the firm moves or closes. Direct dials are tied to a person and go stale when that person changes role. Mobiles are personal data in most jurisdictions, have the highest churn and the strictest consent rules. Ask which type the export contains before comparing price per record.
Phone data also decays faster than email. A work email usually survives a role change at the same company; a direct dial often does not, and a reassigned mobile can reach a stranger within months. Refresh cadence is therefore a better predictor of quality than the size of the database.
| Number type | Typical accuracy | Best use | Main risk |
|---|---|---|---|
| Company switchboard | High while the firm is trading | Routing to a department or a gatekeeper | Slow, and often screened by reception |
| Direct dial | Medium, falls after role changes | Reaching a named person at work | Reassigned after the person moves |
| Mobile | Lower and jurisdiction-dependent | Multi-channel follow-up where consent allows | Personal data rules and high churn |
| Main line from a directory | Medium | Account-level contact | Duplicates across branches and trading names |
The export usually mixes all four without saying so. Ask for the breakdown by type, not just the total record count.
Numbers come from public directories, licensed telecom data, customer contribution and web scraping. The mix matters less than how often it is re-checked. Ask for the refresh date of the rows in your sample, not the database.
Direct dials break when someone leaves. A provider that re-verifies quarterly will still be wrong for anyone who moved last month. For senior roles, expect faster decay than for individual contributors.
Mobile numbers are ported between carriers and recycled after a contract ends. A number can be valid and still belong to a different person than the record says. Phone verification confirms the line is live, not who owns it.
The same data carries different obligations in the UK, the EU and the US. A provider that is compliant for US business calls may not cover the consent basis for German or Dutch contacts.
Email invalid at export is measured only where BounceZero has run a study. Every other accuracy claim below is vendor-reported and should be treated as a range, not a fact.
| Provider | Strongest region | Email invalid at export | Phone coverage | Refresh |
|---|---|---|---|---|
| Apollo | Global SMB and mid-market | 5.8% invalid plus 18.4% catch-all (measured, 100,000-contact study) | Vendor-reported, broad | Vendor-reported, continuous |
| ZoomInfo | US enterprise | About 22% invalid (measured) | Vendor-reported, strongest in the US | Vendor-reported, continuous |
| Cognism | UK and Europe | Not measured by BounceZero; estimates 10 to 25% | Vendor-reported, strong in UK and EU | Vendor-reported |
| Lusha | Global, SMB | Not measured; estimates 10 to 25% | Vendor-reported | Vendor-reported |
| Clay | Global, enrichment-led | Not measured; estimates 10 to 25% | Aggregates other providers | Depends on the source |
Full provider comparison: B2B contact database providers. Measured studies: Apollo email accuracy and ZoomInfo email accuracy.
Not the rows the sales team chose. Same region, same seniority, same company size as the list you intend to build.
Count how many reach the named person, how many reach the company, and how many are dead or wrong. That ratio is the only phone accuracy metric you have.
A bulk verification job gives you invalid, catch-all, unknown and role per row. Email quality and phone quality come from the same sourcing pipeline, so one predicts the other.
Confirm the lawful basis for calls and texts in each country you will contact. The provider selling the data does not carry that risk for you.
Measure how much of the sample changed. A provider with a fast decay rate will look worse in month two than in the first demo.
In the UK, business-to-business email can rely on legitimate interest with a clear opt-out, while cold calls to businesses still need screening against the TPS and CTPS. Germany, Austria and the Netherlands require prior consent for B2B email. In the US, the TCPA governs calls and texts and carries a private right of action. Buying a compliant database does not transfer the obligation to check how you use it. This is a working summary, not legal advice.
Coverage, accuracy, refresh and price compared
How to evaluate a provider
The three routes that work, and their limits
Sources ranked by data quality
Invalid, catch-all and unknown rates in full
Where verification fits in the pipeline
A contact number database is a searchable store of business phone numbers linked to named people and companies, sold as seat-based search access or export credits. Exports usually mix company switchboards, direct dials and mobile numbers, and the mix matters more than the total record count.
Accuracy depends on the number type and the refresh cadence. Switchboards stay accurate while the firm trades, direct dials fall after role changes, and mobiles are recycled. Email accuracy at export, which comes from the same sourcing pipeline, measured 5.8% invalid plus 18.4% catch-all at Apollo and about 22% invalid at ZoomInfo.
Free tiers exist but are capped and usually exclude exports. Use a free tier to test coverage for your segment, then pull a random sample and check it before paying for volume. A free tier is a demo of coverage, not a quality test.
It depends on the country and on whether you are calling a business or an individual. In the UK, business calls still need screening against the TPS and CTPS. In the US, the TCPA governs calls and texts. Buying the data does not carry the consent obligation for you. This is not legal advice.
Export the rows you intend to use and run them through a verifier. You get valid, invalid, catch-all, unknown and role per row, then remove invalid, hold catch-all and unknown out of cold volume, and re-verify before every send. BounceZero includes 100 free checks a month and refunds unknowns.
Upload the export, get valid, invalid, catch-all, unknown and role per row, and plan calls for the numbers and email for the confirmed addresses. 100 free checks a month, unknowns refunded.
Ayoub built BounceZero's 5-stage validation pipeline, its dedicated sending infrastructure, and the high-availability database behind every verification. Previously built high-volume email delivery systems. Trained at 1337 Benguerir (École 42 network, 2019). Open-source: bgp_analyzer.
Apollo, ZoomInfo, Lusha and Cognism all have accuracy gaps - verify exports before sending
Waterfall enrichment, AI first lines, and is $167-446+/mo worth it
Domain search vs ICP prospecting - when to use each and how to stack them
Three routes that work, and their limits
Up to 30% of Apollo emails will bounce without verification
20-25% accuracy gap on ZoomInfo email data
Remove the 18-28% of Lusha contacts that bounce
Continue through related topics