Every email verification vendor on the market today will tell you they hit 99% accuracy. Some say 99.9%. A few brave ones say 99.99%. None of them define what they are measuring, and almost none of them publish how they test it. This is the most important number in our entire category, and the industry has collectively agreed not to look at it too closely.
We think that is silly. So this post is the version of the accuracy conversation we wish we had read before we started building CleanMyList.
Accuracy of what, exactly?
There are three different things people mean when they say a verifier is accurate, and they are all measured differently.
- Bounce match rate — given a list, how well does the verifier predict which addresses will actually bounce when you send to them.
- Catch-all coverage — when a domain accepts every address, can the verifier tell you anything useful about it, or does it shrug and say risky.
- Decision stability — if you run the same address through the same verifier twice, an hour apart, do you get the same answer.
Most marketing numbers are bounce match rate, measured against a curated test set that the vendor controls. That is fine for marketing, but it is not how your list looks. Your list has corporate domains, abandoned consumer mailboxes, and a long tail of weird TLDs that the vendor probably did not test.
How we measure it at CleanMyList
We hold out a rolling sample of every list our customers process. After 30 days, we compare our verdicts against the actual bounce data from the customer's ESP. Then we publish the gap. Every month, on the same page. If we get worse, you will see it before we do.
We also run a private test set built from spam-trap intelligence partners, but we do not lead with the score from that set, because it overstates accuracy by about three points compared to real customer data. Most vendors lead with the easier number. That is the gap we are trying to close in this industry.