When cold email lands in Junk on Microsoft 365 but not Gmail, the answer is usually not "your whole domain is dead." It is usually that Microsoft and Gmail are reading different signals, with different tolerance for cold patterns, at different points in the route.
A blended deliverability score hides the problem. A campaign can show 90 percent placement on Gmail seeds, 60 percent placement on Microsoft 365 seeds, and still look "fine" in a dashboard that averages providers together. That average is operationally useless. Your buyer either uses Gmail, Microsoft 365, Outlook.com, Yahoo, or a gateway in front of one of them.
The short answer
Microsoft 365 is usually harder on cold email because it protects a huge share of enterprise inboxes, combines tenant-level policies with Microsoft Defender filtering, and treats suspicious outbound behavior aggressively when the sender is also Microsoft-hosted. Gmail is strict too, but its public sender guidance gives clearer thresholds to manage.
That difference matters because cold email is not judged only by the text in the message. It is judged by sender authentication, domain age, complaint history, recipient engagement, link behavior, volume changes, mailbox provider reputation, and the specific receiving environment. The same email can look acceptable to Gmail and risky to Microsoft 365.
Do not troubleshoot this from a single seed score. Split the data by provider first: Gmail, Microsoft 365 or Outlook, Yahoo, and secure gateways. Then make one change at a time.
Why Microsoft 365 filters cold email harder
Microsoft documents how seriously it treats suspicious outbound traffic. In Microsoft Defender for Office 365, messages identified as outbound spam can be delivered through a separate high-risk delivery pool instead of the normal outbound IP pool. Microsoft also explains that outbound messages are scanned for spam and that spam can be routed through a lower-reputation pool to protect the rest of the service.
For cold senders, that is the important mental model. Microsoft is not only asking "is this message authenticated?" It is asking "does this sender behavior look like risk?" If you are sending cold traffic from a Microsoft-hosted mailbox into Microsoft-protected organizations, you are putting the campaign in the strictest possible lane.
On the receiving side, many Microsoft 365 tenants also have their own security policies, admin decisions, quarantine rules, and third-party gateways layered in front of the inbox. That means one Microsoft-hosted company may accept your email while another Microsoft-hosted company junks or quarantines the same message. This is why a generic "Outlook deliverability score" is never enough.
Why Gmail can still inbox the same campaign
Gmail is not easy. It is just more legible. Google's sender guidelines tell senders to authenticate mail with SPF or DKIM, avoid impersonation, maintain valid DNS, format messages correctly, and keep spam rates reported in Postmaster Tools below 0.30 percent. For bulk senders, Google also expects DMARC alignment and one-click unsubscribe for commercial or promotional email.
So when Gmail inboxes a cold email that Microsoft 365 junks, the campaign is not automatically "good." Gmail may simply have more positive or neutral signals for that sending domain at that moment. Microsoft may be seeing risk in a different place: the sending mailbox, the receiving tenant, an attachment/link pattern, the message route, or recent behavior from similar senders.
This is also why you need a deliverability recovery playbook before the score breaks. If Microsoft placement falls first, Gmail can lag behind for a few days and make the sender look healthier than it is.
Highest risk for cold campaigns when the recipient side is also Microsoft-heavy.
Can add tenant policy, Defender, quarantine, and security-gateway layers.
Provider-specific reputation issue. Do not average the result away.
Run the exact email from a separate warmed sender and compare provider placement.
Fix one: test by provider, not by average
The first fix is measurement. A seed test is useful only if it shows provider-level placement. If the test combines Gmail, Microsoft 365, Yahoo, and private domains into one score, it will hide the issue you actually need to solve.
Run a test like this:
- Use at least 20 Gmail seeds and 20 Microsoft 365 or Outlook seeds.
- Send the exact production email, not a cleaned-up version for the test.
- Keep the sending mailbox, subject line, body, signature, and tracking settings unchanged.
- Record inbox, spam, quarantine, delay, and missing-delivery separately by provider.
- Change one variable, then run the same test again.
Fix two: separate the Microsoft sender lane
If Microsoft 365 placement is weak and the sender mailbox is Microsoft-hosted, isolate that variable. Send the same test from a separate warmed Google Workspace sender, or from another dedicated sender lane that has clean SPF, DKIM, DMARC, and stable volume history. Then compare Microsoft seed placement again.
This does not mean "always use Gmail forever." It means you should not keep pushing cold traffic through a lane that provider-level tests already show is failing. If Microsoft-hosted senders are consistently worse against Microsoft 365 recipients, move cold traffic elsewhere and reserve Microsoft-hosted mailboxes for warm, transactional, internal, or manually reviewed communication.
If your list is heavy with enterprise domains, run the same pre-send discipline you would use for security-gateway routing: identify the receiving infrastructure, segment high-friction domains, and avoid making one sending lane carry every type of risk.
Fix three: make the email boring enough to pass
Copy does matter, but not in the clever way people hope. The best cold email for a provider test is short, plain, and easy to classify. Gmail and Microsoft both have reasons to distrust noisy email: links, tracking pixels, heavy HTML, styled signatures, attachments, image-only signatures, and sudden volume spikes.
For the first recovery test, use this format:
- 25 to 50 words in the body.
- No link in the body.
- No link in the signature during the test.
- No open tracking pixel.
- No image, table, banner, or styled signature.
- One clear ask in plain text.
This feels extreme until you compare results. A LinkedIn URL in a signature, a tracking domain with weak reputation, or a styled HTML block can be enough to turn a borderline provider into a Junk placement. Strip the message first, then add pieces back only after the plain version passes.
Fix four: make authentication and unsubscribe boring
SPF, DKIM, and DMARC will not save a bad campaign, but missing or misaligned authentication can kill a decent one. Google tells senders to authenticate mail, keep DNS clean, avoid Gmail impersonation, and monitor spam rates in Postmaster Tools. Microsoft similarly treats authentication and suspicious sending behavior as part of its outbound spam controls.
For commercial or promotional sends at scale, one-click unsubscribe also matters. Google's sender FAQ points to List-Unsubscribe headers and RFC 8058 for one-click unsubscribe behavior. RFC 8058 defines the header signal that lets mail clients process a one-click unsubscribe safely. Even if your cold email motion is low volume, the lesson is practical: make opt-out easy, visible, and reliable so recipients do not use "report spam" as the only escape hatch.
What to do this week
If Microsoft 365 is the weak provider, do this before sending more volume:
- Run a provider-split seed test with Gmail and Microsoft 365 seeds separated.
- If Microsoft 365 placement is below 80 percent, pause volume to Microsoft-heavy segments.
- Retest the same message from a separate warmed sender lane.
- Strip links, images, tracking, and styled signatures from the body and retest.
- Confirm SPF and DKIM pass, and confirm DMARC exists and aligns where required.
- Check Google Postmaster Tools for Gmail spam rate and reputation signals.
- Move high-friction enterprise accounts into a manual or alternate outreach path.
The goal is not to make Microsoft 365 "easy." It is to stop treating Gmail placement as proof that every provider is fine.
Questions cold email teams actually ask
Why do my emails go to junk in Outlook but not Gmail?
Outlook and Microsoft 365 can judge the same message differently than Gmail because they use different reputation signals, tenant policies, spam controls, and security layers. If Gmail inboxes while Outlook junks, split your seed test by provider and troubleshoot Microsoft placement separately.
How do I stop emails going to Junk in Outlook 365 as a sender?
Start with provider-specific seed testing, then fix the sender lane. Confirm SPF, DKIM, and DMARC. Remove tracking, links, images, and styled signatures from the test message. If a Microsoft-hosted sender performs poorly against Microsoft 365 seeds, retest from a warmed non-Microsoft sender lane.
Is Microsoft 365 worse than Gmail for cold email deliverability?
Microsoft 365 is often less forgiving for cold email, especially around enterprise tenants and Microsoft-hosted sending lanes. Gmail is also strict, but Google publishes clearer sender requirements and spam-rate thresholds. Treat both as separate providers instead of assuming one score represents the whole market.
Should I remove links and tracking from cold email?
Yes, at least during diagnosis. Links, tracking pixels, branded tracking domains, and HTML signatures can all add risk. Run the first seed test as plain text with no links or tracking. Once that version places well, add one element back at a time and retest.
What is the Microsoft 365 high-risk delivery pool?
Microsoft's high-risk delivery pool is a separate outbound IP pool used for lower-quality messages such as spam or backscatter, according to Microsoft Learn. For cold email teams, it is a reminder that Microsoft treats suspicious outbound behavior as a reputation risk, not just a content issue.
How often should I run cold email seed tests?
Run a baseline seed test before launch, then retest after any major change: new domain, new mailbox provider, new tracking domain, new template, new link, or sudden provider-specific drop. For active campaigns, weekly provider-split checks are enough unless the placement trend breaks.
