Email scrubbing before a send that matters
How the list got this way
- A campaign was scheduled against a list that has not been mailed in months, and nobody knows what state it is in.
- A sending platform sent a warning about the email bounce rate on the last send, and the next one risks a suspension.
- Several lists were merged for a launch, each with a different age and a different collection method.
- A new sending domain or IP has no reputation yet, and the first send decides how the next year goes.
Why the receiving side punishes a bad send
Mailbox providers read a high rejection rate as a signal that the sender does not know who is on their list, and the usual reason for that is purchased or scraped data. The penalty lands on the sending domain and IP rather than on the campaign.
Sending platforms enforce this before the mailbox providers do, because their own reputation is shared across customers. Amazon SES puts an account under review at a 5% email bounce rate and can pause sending entirely at 10%, which takes password resets down alongside the newsletter.
Reputation recovers slowly. A single bad send can cost weeks of degraded placement, which is why scrubbing before the send is worth more than analysis after it.
What a check removes, and what it leaves
Both columns matter. A verifier that only advertises the left one is describing half the job.
| Removed by checking | Still there afterwards |
|---|---|
| The rejections that would have driven the email bounce rate into a platform threshold. | Predict complaints. Someone who forgot signing up will mark it as spam regardless of how deliverable the address is. |
| Dead domains from companies that folded or rebranded since the list was collected. | Resolve catch-all domains, which on a B2B launch list can be a large share of the rows you most want to reach. |
| Role addresses that attract complaints, where you would rather choose than discover. | Detect a pristine spam trap, which is indistinguishable from a real mailbox at the protocol level. |
| Disposable domains, which are pure risk on a reputation-sensitive send. | Make an old list into a warm one. Deliverability and engagement are separate problems. |
What to do instead
Scrub close to the send date
Verdicts age. A file checked three months ago has drifted, and on a reputation-sensitive send the recent check is the one worth paying for.
Send to your engaged segment first
Opens and clicks from people who want the mail establish a pattern before the colder rows arrive. This matters more on a new domain than the scrub does.
Hold the unresolved rows out of the first wave
Catch-all and unknown rows carry unknown risk by definition. Send to them once the run is established and you can watch what comes back in isolation.
Watch the bounce log during the send, not after
If rejections climb past a couple of percent as the send progresses, stop it. The threshold that suspends an account is reached faster than most schedules allow for.
Questions people ask
What email bounce rate is considered safe?
Under 2% is the usual working target, and platforms act well before that becomes catastrophic. Amazon SES reviews accounts at 5% and can pause sending at 10%. Note this is the email bounce rate, meaning rejected messages, and not the web analytics metric of the same name.
How long before a send should a list be scrubbed?
Close to it. Days rather than months. Addresses die continuously and a verdict is a statement about the moment it was taken.
Does scrubbing improve open rates?
Indirectly and only sometimes. Removing dead addresses raises the percentage because the denominator shrinks, and better placement can raise it genuinely. Neither makes uninterested people read your mail.
Can scrubbing get a suspended account reinstated?
It is usually required before a platform will consider it, but on its own it is not enough. They want to see the collection practice changed as well as the list cleaned.
Related list problems
Measure your own list before you clean it
100 free checks a month, no card. The rows we could not get an answer on come back labeled rather than guessed at, and we never bill for them.