The benchmark method, in full
A benchmark is only as useful as its method. Here is ours, fixed and public, so every figure it produces can be checked against the standard it was held to.
The short version
- Hunter tested 3,000 real business addresses across 15 verifiers (published 3 November 2025) and measured overall accuracy of 31.2% to 70.0%. A disclosed field test like that is the standard worth meeting.
- Coverage and accuracy are reported together. Either number alone can be improved by making the product worse.
- Results are broken out by provider, hard ones included. A benchmark that only reports wins is marketing in a lab coat.
- Ground truth is delivery: an address is scored on what happened when mail was sent to it.
The market figures this method answers
Advertised accuracy sits above 99%. Whole-list field tests land between 31.2% and 70.0%. Every figure below is dated and sourced, so you can check it yourself.
| Figure | Value | Source and date |
|---|---|---|
| Advertised accuracy, ZeroBounce | 99.6% | ZeroBounce's own site, read September 2026 |
| Advertised accuracy, Clearout | 99.73% | Clearout's own site, read September 2026 |
| Field test, 15 verifiers on 3,000 business addresses | 31.2% to 70.0% | Hunter, published 3 November 2025 |
| Field test, top ten verifiers | 57.6% to 70.0% | Hunter, published 3 November 2025 |
| Bulk turnaround per 100,000 addresses | 7 minutes (Emailable) to under 2 hours (DeBounce); 45 to 60 minutes at ZeroBounce | Each vendor's own site, read 24 September 2026 |
Vendor figures are the vendors' own claims, quoted as published. Field-test figures count the whole list, catch-alls and unknowns included.
The gap between 99% and 70% is a denominator, not a disagreement about the same thing. You can read why verifiers disagree on the same list, and what accuracy is actually achievable on domains that answer honestly. The accuracy page sets out the two-number argument in full.
What is measured
| Metric | Definition | Why both |
|---|---|---|
| Coverage | Share of the test list that produced a definitive verdict. | Alone it is gameable: resolve every ambiguous address to valid and it hits 100%. |
| Accuracy | Of those verdicts, the share that held when mail was actually sent. | Alone it is gameable the other way: judge less and it approaches 100% on what is left. |
| False positives | Addresses called valid that hard bounced. | The costly error. Reported separately rather than averaged away. |
| False negatives | Addresses called invalid that accepted mail. | The quiet error: customers you deleted. |
| Per provider | The same four figures for Gmail, Microsoft 365, Yahoo, and self-hosted domains. | A single average hides that Yahoo is close to unmeasurable. |
How it runs
Build a list with known outcomes
Each address has a real state, established by sending to it and recording what happened. No other verifier is asked. A verifier scored against a verifier measures agreement, not truth.
Describe the sample honestly
Size, how it was assembled, the mix of providers, and the B2B to consumer split. Any sample has a bias, and the useful move is naming it in the published write-up.
Verify without special handling
The same pipeline a customer's list goes through. No retries we would not normally do, no domains excluded for being difficult.
Send, and record what bounced
The ground truth is delivery, not a second opinion. Every address is scored on whether mail to it was accepted or hard bounced.
Publish every figure
Method, dataset description, date, and per-provider results, hard providers included, so no number appears without the context it was measured in.
How the hardest addresses are scored
Questions
Why publish the method in full?
Because a method you cannot read is a method you cannot check. A fixed, public standard is what makes a figure mean something.
Why score against sends rather than another verifier?
A verifier scored against a verifier measures agreement, not truth. A send either reaches a mailbox or bounces, and that is the answer that matters to your sender reputation.
Why report false positives on their own?
An address called valid that hard bounces costs you reputation. Averaging it into one accuracy figure hides the error you most need to see.
How does this compare with Hunter's benchmark?
Hunter tested 3,000 real business addresses across 15 verifiers and measured overall accuracy of 31.2% to 70.0%, published 3 November 2025. The same test put the top ten between 57.6% and 70.0%. It is the most cited public field test in this market, and the ZapBounce method is built to be read beside it.
What accuracy do email verifiers advertise?
Vendor pages advertise figures above 99%: ZeroBounce publishes 99.6%, Clearout 99.73% and Reoon above 99%. Those figures count only the addresses that reached a verdict, which is why a whole-list field test lands far lower.
Measure us on your own list
100 free checks a month. Run a sample and count the unresolved share.