Accuracy methodology
Every incumbent measures accuracy on the easy case — addresses a simple mailbox check already handles.
That number goes to zero on the hard case: ~15% of B2B addresses sit on catch-all domains (median per-list share, per MailerCheck’s published data).
That 15% is exactly what determines whether verification is worth paying for.
Loading…
Calibration, not marketing
Most verifiers publish a single accuracy number and leave it there. We’d rather show you what we actually tested, on addresses we actually own, and say plainly where the current ceiling is — not round it up.
A controlled domain we own, configured as catch-all, with a known set of real mailboxes and a known set of fake ones — ground truth we control on both sides, not a guess.
Every real mailbox resolved correctly. Every fake one was held at a low, uncertain confidence score rather than wrongly confirmed as real — the model is being deliberately conservative here, not wrong. Measured against a strict pass/fail bar, that’s a 52% pass rate; measured against “did we ever falsely confirm a fake mailbox,” it’s zero.
For this specific case — a catch-all domain on a major enterprise mail platform — a stronger identity signal exists on that platform’s side, but the platform itself rate-limits it aggressively. We reserve that signal for a rarer, more critical failure mode instead of spending it on every catch-all check, protecting reliability for every customer on that path. Expanding its use is a real tradeoff we’re weighing openly, not a bug we’re hiding.
A documented gap you can verify beats an inflated number you can’t. This result feeds directly into our own calibration — not a customer-facing sample used to move a headline.
This reflects one narrow, controlled test case — not overall product accuracy. It doesn’t cover general SMTP verification, non-catch-all domains, or other mailbox providers, all of which are measured separately above.
Vendor-published benchmarks
Not independent third-party audits — Hunter and Allegrow both compete in this space. Numbers are accurately quoted from their own published methodology, cited so you can check them yourself.
Hunter’s own 3,000-email test of 15 verifiers (itself included) — top real-world accuracy score.
ZeroBounce’s catch-all false-positive rate in Allegrow’s own 1,222-email test — missed 30% of real contacts.
Target accuracy by segment
| Segment | Target accuracy |
|---|---|
| Non-catch-all (standard SMTP) | ≥ 99% |
| Enterprise tenant catch-all | ≥ 95% |
| Google Workspace catch-all | 75–85% |
| Generic SMTP catch-all | 65–80% |
| SEG-fronted (Proofpoint/Mimecast) | 60–75% |
Feature comparison
| Feature | ZeroBounce | NeverBounce | Kickbox | BriteVerify | skink |
|---|---|---|---|---|---|
| Catch-all resolution | ❌ extra charge | ❌ unknown | ❌ unknown | ❌ unknown | ✅ resolved out-of-band |
| Enterprise mailbox existence | ❌ | ❌ | ❌ | ❌ | ✅ resolved |
| Google Workspace catch-all | ❌ | ❌ | ❌ | ❌ | ⚠️ probabilistic |
| SEG handling | Silent unknown | Silent unknown | Silent unknown | Silent unknown | ✅ resolved behind the gateway |
| Greylisting correct* | ❌ | ✅ some | ✅ | ⚠️ | ✅ auto-retry (opt-in) |
| Confidence score, not binary | ❌ binary | ❌ binary | ⚠️ | ❌ binary | ✅ confidence score |
| Exposed reasoning | ❌ | ❌ | ❌ | ❌ | ✅ reasons[] + full signals |
| Credits never expire | ✅ | ❌ 12mo | ❌ 12mo | ❌ 12mo | ✅ |
Checked against public vendor docs, July 2026 — not an independent audit.
* Greylisting correct: Greylisting is detected automatically. Set allow_async_followup on your request and we'll retry in the background once the greylist window passes, instead of returning a final verdict on the first attempt.