An email verifier confirms a mailbox. It cannot confirm a person.

A client sent us a file last quarter with a bounce rate they were proud of. Under one per cent. Their verifier had cleared it, their ESP was happy, and their reply rate was still on the floor. They wanted to know what was wrong with their copy.

Nothing was wrong with their copy.

Every email verifier on the market — including ours — answers one question, and answers it well: will this mailbox accept a message? That is a useful question. It is also a much narrower one than most teams realise they are asking.

Take a row that passes every check a verifier runs. Syntax valid. Domain resolves. MX record live. Mailbox exists and accepts mail. The verifier returns valid, and the verifier is not wrong.

The person left in March.

Why the mailbox outlives the person

Think about how your own company handles a leaver. Almost nobody deletes the mailbox on the last day. It gets kept for handover, forwarded to a manager, pointed at a shared inbox, or just left running because deleting it is nobody's job and there is no upside in being the person who breaks something. Six months later the address still accepts mail perfectly well. Nothing in it reaches a human who cares.

From inside your reporting this looks like a good send. Delivered, no bounce, no complaint. Just silence. And every outbound team on earth has been trained to read silence as disinterest — bad timing, wrong message, try again next quarter — rather than as a record that was never real in the first place. That misreading is expensive, and it is completely rational given what the data shows you.

The five states a verifier cannot see

• They left the company. Current role carries an end date.
• They are openly job-hunting. Whatever you are selling, they are not the one who will sponsor it.
• They are not a full-time employee. Freelance, contract, intern, student, retired — on the org chart in name only, and rarely holding budget.
• They work somewhere else entirely. The employer on your file is not the employer on their profile.
• Eleven years in the same seat. Technically reachable; realistically not moving, not buying, and not worth the send.

None of these produce a bounce. All of them produce silence.

What this costs at volume

At ten thousand emails a month this is an annoyance you can absorb. At five million it stops being a data problem and becomes a deliverability problem. Sending reputation is a function of engagement at least as much as of bounces, and a list full of technically valid, humanly dead addresses drags engagement down across every campaign sharing that domain — including the ones aimed at people who would have replied.

Every bad record you buy, you pay for twice. Once at purchase, and again when it drags your sender reputation down with it.

The second cost is the one nobody invoices you for, and the one that takes a quarter to undo.

Checking the person instead

So we stopped treating the inbox as the unit of validation and started treating the person as the unit of validation. Every row gets compared against the prospect's live professional profile at the moment of the run, and every removal comes back with a written reason attached to it. (How we measure whether that is working is a separate page, with the method written out.)

In practice the reason matters more than the verdict. “Invalid” gives you nothing to do on Monday morning. “Left the company, March 2026” tells you to go back to the vendor who sold you that record, and hands you the evidence to do it with.