Why your Microsoft SNDS status turned yellow, and what to do
Monitoring deliverability metrics matters for any business that relies on email, and a sudden change in Microsoft SNDS from green to yellow is alarming precisely because the dashboard gives you a color and no explanation. Here is what the color actually measures, what moves it, and the order to work through when it changes.
What yellow actually means
Microsoft's Smart Network Data Services reports a filter result for each of your IPs, color-coded by the share of recipients whose messages received a spam verdict from Microsoft's filter:
Filter result | Share of recipients with a spam verdict |
|---|---|
Green | Under 10% |
Yellow | 10% to 90% |
Red | Over 90% |
Yellow is a very wide band, which is the first thing to understand about it. An IP at 11% and an IP at 89% both display yellow, and they are not the same situation.
Two clarifications that change how you read it:
The color is a filter verdict, not an inbox placement rate. It tells you what Microsoft's filter decided, not where the message ended up. A recipient's safe list or personal settings can override the verdict after the fact, so your actual placement may be better than the color suggests.
The color is not your complaint rate. This is the most common misreading. Complaints are user actions — someone pressing "mark as junk" — and they live in a separate SNDS column and in your JMRP feed. Complaints are a major input into the reputation that produces the filter verdict, but they are not the thing the color measures. Diagnosing a filter verdict by looking only at complaints will leave you without an explanation a good share of the time.
To locate yourself within the band, compare the affected IP against the others in your range. If your other IPs are green, the yellow one is probably near the 10% floor. If most are red, it is probably near 90%. That comparison is the fastest read you can get without additional tooling.
One counting detail worth knowing: a single message sent to ten recipients counts as ten verdicts. A single send to a large corporate list can move the color on its own, which is often why the change looks sudden and unexplained.
What actually drives the shift
Complaint rate. Microsoft's guidance is to stay below 0.3%, and healthy senders typically run under 0.1%. Sustained complaint rates above that threshold are among the most common causes of a yellow or red filter result. Treat it as a trend across a week rather than a daily number — daily values are noisy enough to produce false alarms.
Spam trap hits. The SNDS trap message column should read zero. Trap hits cause rapid reputation deterioration and are usually a signal about how addresses entered your list rather than about what you sent.
Changes in volume or pattern. A sudden increase in send volume, a new campaign type, or a change in frequency can shift the filter verdict before any complaint data catches up.
Abuse flags. The SNDS Comments column surfaces entries like "JMR Block," "Abuse Reported," virus detection, malware hosting, and open proxy. These are categorically different from a reputation dip and need their own response path.
How to investigate, in order
Check the Comments column first. An abuse flag is a different problem from a reputation problem and changes everything about your response.
Compare against your other IPs to estimate where in the 10-to-90 band you sit.
Pull your JMRP feed and look for a complaint spike tied to a specific send.
Check trap hits. Anything above zero takes priority over complaint tuning.
Review the last two weeks of campaigns for changes in content, frequency, volume, or list source.
Cross-reference Google Postmaster Tools. A problem visible only in SNDS points to something Microsoft-specific. A problem visible in both points at your sending program.
What to do about it
Pause aggressive sending from the affected IP rather than pushing through. Suppress the addresses that complained, pulled from your JMRP feed. Then allow five to seven days for Microsoft to re-evaluate, because the filter result responds to accumulated behavior rather than to a single corrected send.
The urgency is real. Yellow left unaddressed tends to escalate rather than stabilize, and it can become red within a few days. But the fix is almost never a single dramatic change — it is reducing the input that caused it and waiting out the re-evaluation window.
When yellow is not a problem
During a new IP warmup, yellow in the first few weeks is normal. Microsoft has no history for the IP and filters conservatively while it builds one. The signals that warrant concern are a sustained red, or a yellow that persists past the fourth week of warmup.
This is the same principle covered in our guide to IP warmup monitoring: the number matters less than whether it is moving in the direction the ramp predicts.
What changed in 2026
Several SNDS and JMRP mechanics shifted this year, and they affect how you diagnose a yellow status:
JMRP reports are redacted. Complaint reports now include original message headers plus selected authentication headers, with the sender address redacted and the message body no longer attached. Content-level diagnosis now depends on your own correlation IDs — campaign or message identifiers set in custom headers. If you are not stamping those today, start, because without them a complaint spike is hard to trace to a specific send.
JMRP feeds must be linked to SNDS. Unlinked feeds are being phased out, and new feeds must be created from an SNDS-authorized account.
Automated Data Access links expire after 30 days. Scripts built on older persistent links can fail silently. Rotate them monthly.
The portal URL is migrating. Microsoft has signaled the current SNDS address will eventually redirect, without publishing a replacement or a date.
These are reported by deliverability practitioners rather than announced in a Microsoft changelog, so treat the specifics as working knowledge and verify against your own account.
Staying out of yellow
Maintaining a sender reputation is an ongoing discipline rather than a fix. Monitor SNDS weekly rather than reactively, so you see a trend instead of discovering a color. Keep list hygiene and data quality tight, since most complaint and trap problems originate at acquisition rather than at send. And segment by engagement so your least-engaged subscribers are not the ones determining your filter verdict.
If the same situation hits you on the Google side, our Gmail deliverability triage guide covers that provider's version of this problem. For the full picture, our deliverability series goes deeper on authentication, reputation, and monitoring.
A note on the SNDS links above: SNDS requires a free account and IP authorization before it will show you any data. Microsoft's own FAQ sits behind that same sign-in, which is part of why so much of the practical knowledge about reading these dashboards lives in vendor documentation rather than in official docs.
