Email Verification Accuracy: What "99% Accurate" Really Means

Many verifiers advertise 98% or 99% accuracy. Without a definition the number tells you little. Here is what to look for.

What accuracy can mean

• Share of correct labels among the addresses the service was confident about.
• Share of all addresses that got a definite answer. Some services mark many addresses unknown and still claim high accuracy on the rest.
• Bounce rate after cleaning. This is the number that matters to you.

Why no verifier can be 100%

The final answer comes from the recipient's mail server. Some servers accept all mail and bounce later (catch-all), some block verification probes, and some use greylisting that delays the first reply. A verifier can only report what the server tells it, so honest services label those cases risky or unknown. See catch-all domains and greylisting.

A fair test on your own list

1. Take a sample of 2,000 to 5,000 addresses from your real list, not a demo file.
2. Include some you know: your own addresses, a few that bounced before and a few that are clearly invalid.
3. Verify the sample and check that the known ones are labelled correctly.
4. Send to the valid group and measure the bounce rate. Under 2% is a healthy result.
5. Check how many addresses ended up risky or unknown and what the service advises for them.

How to read the result

• Valid: send.
• Invalid: remove.
• Risky or catch-all: send in small batches and watch the bounce rate.
• Unknown: retry later, then decide.

What you can do to improve accuracy

Verify close to sending time, because addresses go stale within months. Re-verify when you import a new source. Use real-time checks at sign-up so bad addresses never enter the list. Use the email error codes to understand the bounces that still happen.

Reading a vendor's accuracy claim

When you see a percentage, ask three questions. What was tested: a clean demo list or a real, mixed one? What counted as correct: only the addresses the tool was sure about, or all of them? And what happens to the uncertain ones: are they hidden inside "valid", or shown as risky or unknown?

Why your own bounce rate is the best test

A claim measures a vendor's sample. Your bounce rate measures your list. After you verify and send to the valid group, a bounce rate under 2% is a healthy result for most senders. If it is higher, look at the bounce messages: some may point to causes the verifier cannot see, such as a mailbox that filled up after the check.

Keep the results fresh

An address that was valid in January may be gone by June. Verification is a snapshot, so the closer to the send, the better. For lists that change slowly, once before each large campaign is usually enough.

A short checklist before you trust any number

• Ask how the test list was built and whether it was real data.
• Ask how uncertain results are counted.
• Run your own sample of 2,000 or more addresses.
• Send to the valid group and measure the bounce rate.
• Repeat the test a few months later, because servers and lists change.
• Keep the report, so you can compare vendors on equal terms.

Ask what happens to the "unknown" bucket

Every verifier has addresses it cannot settle. A good one reports them honestly and tells you how to handle them; a poor one hides them inside "valid". Ask for the share of unknowns on a typical list and for advice on sending to them.

Treat accuracy as a conversation

A vendor that explains its limits is more trustworthy than one that promises perfection. Ask for the explanation, test on your own list and keep the one whose answers match what you see in your bounce rate.