Delegate one mailbox
You admin-consent a scoped app and hand over a single probe mailbox. An application access policy locks the grant to that one mailbox. Revoke it in one click.
Trace says Delivered. The inbox is empty. We tell you which filter ate it and how to get it back, read straight off that filter’s own console. No simulation.
Preliminary verdict in ~15 min, final at 24 h. One delegated mailbox is the whole footprint.
Your mail runs the gauntlet of the filters that actually block it
Receiving tenants run Microsoft 365 (EOP plus Strict on Defender P2), Google Workspace, Mimecast, Proofpoint, Barracuda, Cisco, and Sophos — with Spamhaus checked in the same pass. These are the filters your mail runs, not a customer list.
Think VirusTotal, for email delivery. Most tools simulate a filter and guess the outcome. We run your mail through the real one and read the verdict off its console.
You admin-consent a scoped app and hand over a single probe mailbox. An application access policy locks the grant to that one mailbox. Revoke it in one click.
We compose your real mail and send it from your own Exchange Online — same IPs, same DKIM your users carry. It runs every filter on the wall above, from EOP and Strict Defender to Mimecast and Proofpoint.
Because we own every receiving console, the verdict is the real one, with its reason code. Then the part other tools skip: the exact form to file and what to attach.
Some filters accept your message with a 250 OK, then throw it away once the connection closes. Your trace reads Delivered. Nobody gets a bounce. The recipient never sees a thing.
From either end alone, this failure is invisible. Your logs say success. Their inbox says the message never existed. You need someone standing on both ends of the wire at once, which is exactly what a probe is. We sent it from your tenant. We run the tenant it landed in. Empty mailbox, nothing in quarantine at 24 hours? That is the verdict, and we issue it.
Filters do this on purpose, to deny spammers a signal. It denies you one too, which is why these incidents drag on for weeks.
The failures that hurt most are the ones that never throw an error.
Delivered to the inbox.
Delivered, foldered as spam. Reason codes attached.
Held in admin quarantine with a category: spam, phish, malware.
Refused at the edge. SMTP code and NDR classified.
Queued or greylisted. The sender is still retrying. Not a block.
Accepted, then destroyed after the connection closed. No bounce anywhere. This is the one a trace can’t see.
A placement score of 62 doesn’t tell you what to do next Monday. “550 5.7.606, file the Microsoft delist form and attach these headers” does. Every finding ends with the part deliverability tools leave out.
Which layer took it. EOP edge, content filter, Defender detonation, Gmail classifier. Blame gets an address.
550 5.7.606 is an IP-reputation block, not a content problem. Chase the wrong one and you lose a week.
Which portal or vendor form to file, what to attach, what turnaround to expect, and what to do when the first try gets ignored.
M365 shares outbound IP pools, so IP findings are pool-level. We split what you can fix from what belongs to a stranger.
Half of delivery incidents are self-inflicted DNS wounds. We check the boring things first, so the playbook never sends you to a delist form when the bug is in your zone file. On the Direct plan these run continuously.
finding 03 · probe 7f2c…e1 verdict: rejected @ Microsoft EOP edge reason: 550 5.7.606 — sending IP on Microsoft block list action: submit sender.office.com delist (attach NDR + probe ID); ETA 24–48 h; if silent after 72 h → escalation path 2, documented below
Written to be forwarded, unedited, to a vendor or a recipient’s IT contact.
On monitoring, findings don’t wait for a bounce. When a leading indicator trips, the agent posts to your Teams channel while probes still land — with the fix already drafted.
Listing appeared 08:47 UTC; caught it on the 08:59 sweep. I ran fresh probes at 09:04 — still inbox across the gauntlet. Mimecast and Proofpoint weight this list heavily. If the listing stands, expect rejected there first.
Two plans. One if you run mail for a single company, one if you run it for many. Talk to us and we scope it to your situation.
For one company running its own mail.
For teams running mail for many companies.
We scope every engagement to your tenant before any work starts. Nothing gets installed, your MX stays untouched, and the grant never reaches past one mailbox.
Tell us what’s failing. Recipient domain, and any NDR text you have, codes and all. The more of the bounce we can read, the faster the answer.
We respond within one business day.