One mailbox, 5,000 companies, blocked in five days: why your cold emails get blocked by Microsoft

Sunny Ray · Feb 24, 2026 · 5 min read

Warmup cannot save a mailbox you are using like a megaphone.

On February 24 the founder of a staffing and outsourcing group opened our call with a challenge I respect. They do lead generation across all their companies. They had just run a campaign that failed completely and they did not understand why. If we were going to do the same thing they had already done, there was no point hiring us:

But again, if it's just repeating what we already messed up, there's not much value in that.

Fair. So his marketing lead walked us through it, step by step. It is the cleanest case study of why cold emails get blocked by Microsoft I have heard, and every mistake in it is common.

The setup that failed: 5,000 companies, one mailbox

They were launching an AI hiring marketplace for funded AI startups, beta due the following month. They needed a name. The founder had a play that had worked for him before: email people three versions of a name or logo and ask which they prefer. People get emotionally attached to aesthetic choices, then want to know more about the company. Feedback and lead generation in one email.

So they scraped about 5,000 ICP companies from Apollo and started emailing them for feedback on the shortlisted names. Through the mailbox. The one mailbox.

When we initially tried Apollo to send all the email campaigns within the first five days, our email got blocked.

Blocked in five days, then Microsoft

Their own diagnosis was honest and correct:

Most probably it was because we were trying to send large volumes of email through the mailbox.

Then they did what almost everyone does next. They bought a warmup tool to improve deliverability. It did not work. Eventually Microsoft blocked the emails.

Someone on our side said what I was thinking: you failed at a critical step that most people never get past. That is not an insult. It is where nearly every in-house cold email attempt dies. Not at copy. Not at targeting. At infrastructure.

I had described the same trap to another founder two weeks earlier. If you hear about cold email and go do it yourself from your own inbox, you can maybe send five or ten a day. Push past that and you land in spam, lose hope, and decide cold email does not work. It works. Sending it from one mailbox does not.

Why warmup did not help

Warmup tools send and receive friendly emails between mailboxes to build a sending history. They help a new mailbox look like a normal person's inbox. They do not make an abnormal sending pattern look normal.

A single human mailbox that suddenly sends hundreds of nearly identical emails a day to strangers is abnormal, no matter how many friendly emails it exchanged last week. Mailbox providers see the volume spike, the low reply ratio, the recipients who never wrote to you before. Microsoft does not care that your warmup score is high. It cares what your mailbox is doing today.

Warmup is a seatbelt. It does not help if you drive into a wall.

The other mistake: asking strangers to name your product

There is a second problem hiding behind the deliverability one. The ask.

A naming survey works brilliantly with people who already know you: customers, a newsletter list, your network. They have a reason to care. Sent cold to 5,000 companies, the same survey is a stranger asking for free work. Low replies, low engagement, more spam complaints, and those signals feed straight back into the mailbox's reputation.

A bad offer and bad infrastructure compound. Each makes the other worse. If the first email had been an invitation that gave something, replies would have been higher and the mailbox healthier, even before fixing the volume.

What survives: many mailboxes, low volume each, owned by you

The same afternoon, on another client call, Charles described deliverability as the primary challenge in B2B lead generation, the place most companies fail before they ever find out whether the market wants their product. He also said Apollo had caused deliverability problems for us at scale, which is why we moved sending to our own stack.

The setup that survives looks like this.

• Never send cold from your main domain. Buy a handful of lookalike domains so your real one is never at risk.
• Spread volume across many mailboxes. Our rule of thumb, as I described it to a founder earlier that month: each mailbox never sends more than about 10 emails a day, across roughly 100 mailboxes.
• Warm up new mailboxes before they send, and keep warmup running. It is necessary, just not sufficient.
• Plain text, no links in the first email, an offer people want to reply to.
• The client owns every domain and mailbox. If you stop working with an agency, your infrastructure stays with you.

Do the math on that founder's campaign. 5,000 companies at 10 emails per mailbox per day is 500 mailbox-days. With 50 mailboxes, that is ten days. With one mailbox, it is more than a year, or five days and a block. Volume is not the problem. Volume per mailbox is.

If you are already blocked

• Stop sending from that mailbox. Do not keep pushing volume to test whether the block lifted.
• If it was your primary domain, move all cold sending off it permanently.
• Rebuild on separate domains, warm them, and start slow.
• Fix the offer before you relaunch. A better ask lowers complaints, which protects the new mailboxes.

The founder's standard was right. Do not hire anyone who will repeat your mistake with more mailboxes. Hire someone who can show you why it broke. See how mailboxes are set up.

Frequently asked questions