Is a warmup score of 100 accurate? Not as a delivery signal

Sunny Ray · Aug 17, 2026 · 7 min read

A warmup score of 100 is accurate about one thing: how your mailbox is doing inside the warmup pool. It is not a measure of whether your real campaign emails are reaching real inboxes. We had every mailbox in a workspace showing a perfect score while some of those same mailboxes were bouncing at 64% in live campaigns.

So treat the warmup score as a hygiene check, not a health check. To know whether your email is landing, you have to look at what happens to real sends: bounces grouped by sending domain, the mail host behind each domain, and whether replies are coming in at the rate they used to.

What a warmup score measures

Warmup tools work by having your mailbox exchange emails with a network of other mailboxes in the same program. Those mailboxes open your messages, reply to them, pull them out of spam. Over time that activity builds a sending history, and the tool turns the results into a score.

That's useful. A new mailbox with no history looks suspicious to mail providers, and a slow ramp of real-looking traffic helps. I'm not against warmup. We use it.

But look at what the score is built from. It's the behaviour of mailboxes inside the program, talking to each other. It says your mailbox gets along well with the warmup pool. It doesn't say anything about a prospect's mail server at a company you've never emailed, and it can't see a rejection that happens somewhere else in your sending path.

The 346 perfect scores

On August 15 we were chasing a bounce spike in one of our workspaces. Overall bounces had hit 19.3%, 428 bounces on 2,220 sends. Before we found the cause, we checked the obvious dashboard signals.

All 346 mailboxes reported status active. All 346 showed warmup at 100. Including mailboxes that were bouncing at 64% in live campaigns.

If you'd only looked at the mailbox screen, you'd have concluded everything was healthy. Green across the board. Meanwhile a big chunk of the real email going out from some of those mailboxes was being refused.

The cause, when we found it, was one mail host. Every sending domain on that host was bouncing between 26% and 100%, and every domain on the other host was at 0%. The warmup score couldn't see any of it. The warmup traffic and the live campaign traffic weren't failing in the same way, so the score stayed perfect while campaigns broke.

Why a perfect score can coexist with a broken mailbox

There are a few reasons this happens, and they're worth understanding so you stop trusting the number too much.

The audience is different. Warmup mail goes to cooperative mailboxes that are designed to engage. Campaign mail goes to strangers' servers with their own filters and rules.

The failure can be upstream. If something in your sending path rejects mail, the sending tool may record it as a bounce on the campaign side. That's a campaign metric, not a warmup metric.

The score is an average over time. Even when something breaks, a score built on weeks of good warmup history moves slowly, if at all.

None of this means the warmup tool is lying. It's answering a narrower question than the one you're asking.

Signals that do track delivery

If the warmup score isn't the health signal, what is? These are the ones I watch now.

Bounce rate by sending domain, inside one campaign. Same copy and same leads across all mailboxes, so any spread between domains points to the sending side. This is the single most useful check we have.

The MX record behind each sending domain. If the domains that bounce share a mail host and the clean ones don't, you've found your problem.

Reply rate over time, per campaign. A sudden drop in replies on a campaign that was working, without any change in copy or list, is a delivery warning even if bounces look normal.

Real-inbox tests. Send to a few addresses you control on different providers and see where the email lands. It's crude, but it's real.

All of these come from actual sends. That's the point. Delivery is about what happens to real email, so the signal has to come from real email.

How we check health now

The diagnosis method we settled on after August 15 is simple enough to run every week. Pull the contacted and bounced leads for a campaign. Read which mailbox sent each one. Group by sending domain. Compare against the MX record for each domain.

If the bounce rate is flat across domains, the sending side is fine and you look at the list or the copy. If it's split by host, you pause the affected mailboxes and keep the healthy ones sending. On August 15 that meant pausing 91 mailboxes on the dead stack and letting 11 campaigns resume clean on 208 healthy ones.

We still look at warmup scores. A mailbox whose score drops is worth a look. But a score of 100 isn't a reason to stop looking.

What warmup is still good for

I don't want anyone to read this and switch warmup off. It still does its job, as long as you're clear about what that job is.

It gives a new mailbox a history before it carries real campaigns. A brand new address that suddenly sends a lot of email to strangers looks bad to mail providers. Warmup softens that.

It catches some problems early. If a mailbox's score drops, something is off, and it's worth a look before campaigns suffer.

It keeps mailboxes active between campaigns, so they don't go cold.

What it doesn't do is certify that your real email reaches real inboxes. Keep it running, and stop reading a perfect score as an all-clear.

What this means if you're running outbound yourself

If you're a founder running your own sending, this is the practical version.

Don't wait for the warmup dashboard to tell you something's wrong. It may never turn red.

Look at bounces per sending domain at least weekly, and every time the overall bounce rate moves.

Know which mail host each of your sending domains uses. Write it down when you set them up. When something breaks, you'll want that list.

Keep your main domain out of it entirely. Cold outreach should go from separate sending domains, so when one stack fails it doesn't touch the address your customers and investors write to.

And if you'd rather not do any of this yourself, that's a perfectly reasonable choice. Monitoring sending infrastructure isn't the best use of a founder's week.

The broader lesson

Every tool in outbound gives you a score. Warmup scores, verification statuses, deliverability ratings. They're useful because they compress a lot of activity into one number. They're dangerous for the same reason.

When a number says everything is fine and your results say otherwise, believe your results. Then go find the data underneath the number. Nine times out of ten the real story is sitting in a spreadsheet you haven't built yet.

We built our plans so founders don't have to babysit mailboxes. On Growth and Team, sending runs on our warmed mailboxes, and the monitoring is our problem, not yours.

Frequently asked questions