Count Out-of-Office Replies When You Judge a Cold Email Domain
TL;DR: To decide whether a sending domain still reaches the inbox, divide all replies, out-of-office replies included, by emails sent, wait for at least 200 sends, and cut the domain under 0.7%. An out-of-office reply only fires when the email landed in an inbox, so it is evidence of placement. Between 2026-09-01 and 2026-10-14 two domains Cedric Lauster runs in the same campaign scored 100% on Google seed tests; one had 7 replies on 528 sends, the other 0 on 529.
Key Takeaways
- Judge a sending domain on all replies divided by emails sent, out-of-office replies included, because an automatic reply only fires when the email reached an inbox.
- Cut any domain with over 200 emails sent and under a 0.7% overall reply rate, checked every Friday.
- Two domains in Cedric Lauster's fleet both scored 100% on Google seed tests over the same weeks; one had 7 replies on 528 sends, the other 0 replies on 529.
- Under 200 sends is too early to judge, and a domain with a few replies but under 0.7% deserves a second look at 400 sends before it is cut.
- Only judge domains that were actually sending at full volume; an idle domain has no reply rate.
To decide whether a cold email domain still reaches the inbox, count every reply it gets, out-of-office replies included, and divide by the emails it sent. Wait until it has sent at least 200 emails, then cut it if the rate is under 0.7%.
It is the deliverability check I trust most. Counting the out-of-office replies is the part most people leave out, and it is what makes the rule usable on a small fleet.
Why should out-of-office replies count?
Because an automatic reply only fires when your email landed in an inbox.
A mail server does not send "I'm out until Monday" for an email it filed in spam. So every out-of-office reply is a small piece of proof that the domain reached a real person's inbox. Some senders even build a placement test on this: they send to a list of addresses known to reply automatically and count how many replies come back.
The alternative, counting only human replies, does not work at domain level. A domain sends a few hundred emails a week. Human replies on that volume come in ones and twos, and one good or bad week swings the rate wildly. You would be judging a domain on noise.
All replies together give you a number large enough to read.
What does the rule look like in practice?
I run it every Friday, per domain: any domain with over 200 emails sent and under a 0.7% overall reply rate comes out of campaigns.
0.7% of 200 is 1.4. So a domain that has sent 200 emails needs at least 2 replies to pass.
Here is why I trust it more than seed tests. Between 2026-09-01 and 2026-10-14 two of my domains ran in the same campaign, with the same copy and the same list:
| Domain A | Domain B | |
|---|---|---|
| Google seed tests | 190 of 190 in the inbox | 194 of 194 in the inbox |
| Emails sent | 528 | 529 |
| Replies | 7 | 0 |
| Reply rate | 1.33% | 0% |
The seed tests could not tell them apart. Both read a perfect 100% for weeks. The reply count could: one domain was reaching people and the other was not.
Seed inboxes are test accounts. They never mark anything as spam and do not behave like real recipients, so a seed test shows a direction, not a verdict. I still run them every weekday, and I wrote about how I read a single bad seed test. But when seeds and replies disagree, the replies win.
Why wait for 200 sends?
Because small numbers lie in both directions.
Take a healthy domain that really replies at 1.5%. On 200 sends you expect 3 replies. But replies arrive at random, and the chance that this healthy domain shows 1 reply or fewer on 200 sends is about 20%. So one time in five, a good domain would fail the rule by bad luck.
That is why I treat the rule in two steps:
- Zero replies on 200 or more sends: the domain is burnt. A domain that really replies at 1.5% shows zero replies on 200 sends only about 5% of the time.
- One or more replies but under 0.7%: judge it again at 400 sends. At 400 sends, the chance that a healthy 1.5% domain still reads under 0.7% drops to about 6%.
Under 200 sends, I do not judge a domain on replies at all. It is too early.
What do you need to check before you apply it?
That the domain was actually sending.
A reply rate only means something if the domain sent at its normal volume. A domain that ran out of leads, sat in a paused campaign, or was still ramping up has a tiny or empty denominator. It can look dead when it simply was not used. Before judging any domain, compare its sends against what it should have sent that week.
Two more things:
- Judge per domain, not per campaign. A campaign average hides the one domain that went quiet. In the table above, both domains sat in the same campaign, and only a per-domain view shows that one of them had stopped landing.
- Use replies over sends, not over leads. Some tools report reply rate per unique lead. Smartlead's domain report does this, for example. On a multi-step sequence that inflates the rate, because each lead receives several emails. Divide the reply count by the sent count yourself.
Where does this sit next to warmup and seed tests?
It is the third of three checks I run, and the only one that comes from real recipients:
- Warmup reputation. Cheapest check, runs on every mailbox. I covered why the warmup inbox count can hide spam.
- Seed placement tests. Directional, read on the Google seeds, never acted on after one bad day.
- Reply rate per domain at full volume. All replies over sends, 200 sends minimum, under 0.7% is out.
If a domain passes the first two and fails the third, it goes. If your reply rate is falling across every domain at once, the problem is probably not one domain at all, and this is the order I use to diagnose a reply rate drop.
How do you run this on your own domains this week?
- Pull sends and replies per sending domain for the last two weeks. Replies of any kind, out-of-office included.
- Drop every domain under 200 sends. Too early.
- Check the remaining domains actually sent at their normal volume.
- Divide replies by sends for each one.
- Zero replies: pull the domain from campaigns.
- Under 0.7% with a few replies: mark it and judge it again once it reaches 400 sends.
- Swap a warmed spare into every slot you emptied.
What is the next step if your domains do not add up?
Start with the domains that score well on seed tests and get no replies. They are the ones a dashboard hides. If you want a second pair of eyes on your numbers, book a call with me and we will go through them domain by domain.
Frequently Asked Questions
Should out-of-office replies count toward a domain's reply rate?
Yes, when you are judging deliverability. An automatic reply only fires when the email reached an inbox, so it is evidence of placement. Cedric Lauster counts all replies, out-of-office included, divided by emails sent.
What reply rate should make me pull a sending domain?
Cedric Lauster pulls any domain with over 200 emails sent and under a 0.7% overall reply rate, checked every Friday. 0.7% of 200 sends means a domain needs at least 2 replies to pass.
How many emails should a domain send before I judge it on replies?
At least 200. On fewer sends, chance alone can make a healthy domain look dead. Cedric Lauster treats zero replies on 200 or more sends as burnt, and judges a domain with a few replies under 0.7% again at 400 sends.
Can a domain pass seed tests and still be in spam?
Yes. Two domains in Cedric Lauster's fleet both scored 100% on Google seed tests over the same weeks, in the same campaign. One had 7 replies on 528 sends, the other had 0 replies on 529.
Should I calculate reply rate per lead or per email sent?
Per email sent. Reply rate per unique lead inflates the number on a multi-step sequence, because each lead receives several emails.
What if a domain was not sending at full volume?
Do not judge it on replies. A domain that ran out of leads, sat in a paused campaign or was still ramping up can look dead simply because it was not used. Check its sends first.
