Unknown Email Verification Results: What They Really Cost

Unknowns aren't free. Many verification tools don't charge you for unknown results, so it's tempting to treat them as free. They aren't. Your infrastructure, subscription and time are fixed costs spread across the addresses you actually classify. Every unknown is a contact you paid to process and can't use.

This post covers what causes unknowns, how to measure what they cost, and where infrastructure (including proxies) helps and where it doesn't.

01What an unknown result actually means

"Unknown" means the verifier couldn't reach a confident answer: valid or invalid. It isn't a verdict on the address. Three causes account for most of it:

• Catch-all domains. The mail server accepts every address, so SMTP can't confirm whether a specific mailbox exists. No infrastructure fixes this — it's a limit of the method.
• Connection timeouts. The server didn't respond in time, often because of throttling, greylisting or network issues. These are often temporary.
• SMTP policy rejections. The server refuses the check itself, typically with a code like 5.7.1, because it distrusts the connecting IP. This is a reputation problem, not a data problem.

02How to measure the cost

The metric that matters is cost per classified result:

Here's an illustrative example. Say you spend $1,000 on infrastructure and tooling to verify 100,000 addresses:

At 5%, the cost difference is small — about 4% more per usable result than at 1%. If your unknown rate is already low, don't over-engineer this. The metric starts to hurt at high unknown rates or very large volumes.

03What we see in our own outbound work

In our own outbound campaigns, unknown results sit at around 5% of addresses. Two observations from running it day to day:

• Retries recover some of them. When we retry a batch of unknowns, roughly 30% come back with a definite answer. That's an indication from our workload, not a guarantee, but it's enough to make one retry worthwhile. Retrying catch-all results is wasted effort, so separate unknowns by cause before you retry.
• IP reputation moves the number. When connections are rejected because the server distrusts the IP, a cleaner IP reduces unknowns. In our experience, moving to a clean-reputation proxy setup can take the unknown rate from about 5% to about 1%. Your results will depend on your list, your volume and your current setup.

04Where proxies help, and where they don't

• Policy rejections and some timeouts, because those depend on how the receiving server sees your IP. A clean, B2B-only IP pool avoids the reputation problems that shared or residential ranges tend to carry.

• Catch-all domains. If a server accepts everything, no IP changes the answer. Anyone who tells you a proxy eliminates unknowns is overselling it.

For what it's worth, Proxy25's IPs are clean-reputation and B2B only, with no residential IPs. That's relevant if your unknowns are mostly policy rejections, and irrelevant if they're mostly catch-all.

05How to investigate a high unknown rate

1. Split unknowns by cause: catch-all, timeout, policy rejection.
2. Retry timeouts once. Don't retry catch-alls.
3. Check policy rejections against your IP. If the codes point at reputation, test a cleaner setup.
4. Compare setups under identical conditions: the same list, same time window, same retry logic.
5. Track three numbers: classification accuracy, connection reliability and cost per classified result.

06The takeaway

A high unknown rate is a prompt to investigate, not proof that your data or infrastructure is bad. Find out which cause is behind it, retry what's retryable, and judge any setup on cost per classified result and accuracy together. If your unknowns are driven by IP reputation, Proxy25 is one option to test against your own workload.