You will see a figure quoted for annual list decay in every article on this subject, usually somewhere between 20% and 30%, usually with no source attached. It circulates because it is convenient rather than because anyone measured your list.
The mechanism is more useful than the number. Addresses do not expire on a schedule. They stop working for specific reasons, and those reasons have very different rates depending on who is on your list.
The largest cause on a business list is people changing jobs. The address belongs to the employer and it goes when the person does, usually with no forwarding and no notice. That rate follows employment turnover in the sector rather than anything about email.
On consumer lists the causes differ. ISP mailboxes die with the broadband contract, which is why comcast.net and orange.fr addresses decay in a way gmail.com addresses do not. Free webmail accounts are rarely deleted, so they stay technically valid while nobody reads them, which is a reachability problem rather than a validity one.
What this means for re-verification
B2B lists: before any large send if the data is over six months old. Employment turnover is the driver, and in high-churn sectors it is faster than that.
Consumer lists: annually is usually enough, and before any campaign following a long gap.
ISP-heavy consumer lists: more often, because the mailbox dies with the contract and nothing tells you.
Free webmail: validity holds up well and engagement does not. Verification will keep saying valid while the address has been unread since 2016, so your own open data is the better signal.
Where this argument costs us something
The short version
- Set re-verification frequency by list type rather than by a general rule.
- Treat a B2B list as decaying roughly twice as fast as a consumer one.
- Use engagement data for free webmail, where validity outlives usefulness by years.
Questions people ask
How often should I re-verify?
Before any large send to data over six months old for B2B, annually for consumer lists, and whenever a list has been idle for a while. Frequency should follow the decay mechanism rather than a calendar rule.
Do valid addresses become invalid gradually?
No, they change state abruptly. Someone leaves a job, a contract ends, an account is closed. Decay looks gradual in aggregate and is a series of discrete events underneath.