Email verification accuracy: how it is measured and why it matters

Email verification accuracy is a measure of how correctly a validator determines the status of an address. An ideal service would classify every address precisely: “exists,” “does not exist,” “risky.” In practice, 100% accuracy is not achievable, so understanding the error metrics is critical both for choosing a tool and for interpreting its results.

Two types of errors

False positive. The validator marks an address as “valid,” but the mailbox does not actually exist. You send a campaign, and the message bounces. Bounces damage sender reputation, so this is the error that most directly hurts deliverability.

False negative. The validator marks an address as “invalid,” but the mailbox does exist. You skip a real subscriber. Reputation stays intact, but you lose audience.

The two error types pull against each other. An aggressive validator that flags anything uncertain as invalid will produce few false positives but many false negatives. A permissive one lets more addresses through and delivers more bounces.

Accuracy metrics

Accuracy (overall). The percentage of correct answers out of all checks. If a validator correctly classifies 9,700 of 10,000 addresses, accuracy is 97%. Clear enough — but it can mislead. If 95% of a list is already valid, a validator that marks everything “valid” scores 95% accuracy without doing any real work.

Precision. Of all addresses marked as “valid,” what share actually is? Precision = True Positives / (True Positives + False Positives). High precision means: when the validator says “send,” you can trust it.

Recall. Of all addresses that genuinely exist, what share did the validator identify correctly? Recall = True Positives / (True Positives + False Negatives). High recall means the validator is not discarding live addresses.

False positive rate (FPR). The percentage of invalid addresses incorrectly marked as valid. This is the most important metric for email marketers: FPR determines how many bounces you will collect after sending to a “verified” list.

What affects validator accuracy

Accept-all domains. Accept-all servers respond “OK” to any address. The validator cannot tell whether a specific mailbox exists. If the validator reports accept-all addresses as “valid” without qualification, it inflates the false positive rate.

Greylisting. Servers that implement greylisting return a temporary 4xx rejection on the first connection attempt. If the validator does not retry, it may incorrectly flag the address as invalid, producing a false negative.

Rate limiting. Gmail and other large providers throttle SMTP connections. During bulk verification, the validator may receive temporary rejections unrelated to mailbox existence.

Temporary outages. A server may be temporarily down. If the check happens at that moment, the result is “unknown” or a false negative. A quality validator retries with a delay.

How to compare validation services

Marketing claims of “99% accuracy” are meaningless without context. 99% of what? Accuracy? Precision? Measured on which sample? On a clean list, any validator looks great. The question is how it performs on messy real-world data.

For an honest comparison: build a test list of 1,000+ addresses where you know the true status of each one — confirmed-existing, confirmed-dead, disposable, accept-all. Run several validators against it and compare their outputs side by side.

Pay attention to how each service handles uncertainty. How many addresses land in “unknown” or “risky”? A good validator says “I do not know” rather than guessing. A large unknown bucket is not a sign of poor quality — it means the service is not passing off guesses as facts.

Practical impact

Consider a list of 100,000 addresses. A validator with 2% FPR lets 2,000 invalid addresses through. You send to them and collect 2,000 hard bounces, pushing bounce rate to 2% — the threshold where many ESPs flag or suspend an account.

A validator with 0.5% FPR lets 500 through. Bounce rate lands at 0.5%, well within safe limits. The difference between 2% and 0.5% FPR can be the difference between a suspended account and stable deliverability.

On the other side, a validator with a high false negative rate discards live addresses. At 3% FNR on a 100,000 list, you lose 3,000 subscribers. If average subscriber value is $1, that is $3,000 in lost revenue.

In both cases accuracy has a direct financial cost. Picking a validator is not a question of “which is cheaper” — it is a question of “which makes fewer errors on my data.”