What is Microsoft SNDS?

The reporting is per IP rather than per domain, which is the reverse of Google's approach and reflects how Microsoft's filtering weighs its inputs.

It used to show something nobody else did: trap hit counts. Microsoft removed them from the data report on July 22, 2026, to protect its anti-abuse systems, and moved SNDS to a new portal. Older guides still tell you to watch that column.

Colour-coded ranges summarize each IP's standing, and a red band correlates tightly with mail landing in the junk folder across Outlook, Hotmail and Live.

JMRP is the companion program and the one that acts. Register there and Microsoft forwards each individual complaint so your platform can suppress the address immediately.

Say you run a dedicated IP, and yesterday's row shows 48,200 RCPT commands, 46,900 DATA commands and 46,850 message recipients. Your filter result is yellow and the complaint rate is 0.4%. (Before July 22, 2026 the row also carried a trap hits count. Microsoft removed it, so a guide that tells you to read that column is out of date.)

The first two numbers are the ones people skip. Your servers asked Microsoft to accept 48,200 recipients and got as far as sending message data for about 46,900. The 1,300 in between were mostly recipients Microsoft refused, which is dead addresses for the most part. That's a bounce-rate reading straight from the receiver's own logs.

Yellow means Microsoft's filters judged somewhere between 10% and 90% of that day's mail to be spam. It's a wide band, with green under 10% and red over 90%. A 0.4% complaint rate is high, and it is now the clearest list-quality reading SNDS gives you, since the trap count that used to sit beside it is gone. Traps still exist: they are addresses Microsoft operates purely to catch senders with poor list sources.

It's tempting to read trap hits as a to-do item for your verifier, and you should be careful with that. A trap mailbox accepts mail like any other, so at the SMTP level it looks valid, and no verification service can reliably pick pristine traps out of a list. Anyone promising otherwise is guessing.

What does bring the number down is dealing with where those addresses came from. Traps get onto lists through purchased data, scraped addresses, and old contacts who were never removed after years of silence. Send each list source on a different day and watch which one the complaint rate and the filter result follow. Then cut or re-permission that source. A sunset rule takes care of the recycled kind over time.

Two practical notes will save you some confusion. SNDS only shows data for days when the IP sent more than about 100 messages to Microsoft's consumer domains, so a quiet IP looks blank. And if you're on a platform's shared pool, ask your provider what their SNDS view looks like for your pool.