What’s the difference between MTA-level and inbox-level filtering?

Still have a question, spotted an error, or have a better explanation or a source we should cite?

Your email just bounced. But was it rejected at the door, or did it sneak inside and get buried in a spam folder? The answer changes everything about how you diagnose and fix the problem.

These are the two filtering stages every message passes through, and they work very differently.

MTA-level filtering happens during the SMTP handshake, before the receiving server formally accepts your message. Think of it as the security check at the door. The receiving MTA looks at things like your IP reputation, whether your domain passes SPF and DKIM, and whether your sending infrastructure looks legitimate. If something fails that check, the server refuses the message outright and sends back a bounce code. You know it failed. Your ESP knows it failed. There's a paper trail.

Inbox-level filtering kicks in after the server has already accepted the message. Now it decides where to put it. The main inbox, the spam folder, a category tab like Promotions, or a custom folder set up by the recipient. At this stage, the filters look at content signals, engagement history (does this sender's mail usually get opened?), and user behavior patterns. Gmail does a lot of heavy lifting here with its machine learning models. So does Outlook.

Here's the part that trips senders up. A spam folder placement at inbox level is silent. Your ESP reports the message as delivered because technically, it was. The receiving server accepted it. From your sending infrastructure's perspective, everything looks fine. But your subscriber never sees it.

That gap is why bounces and spam placement require completely different fixes. A bounce tells you the problem is upstream, usually IP reputation, authentication, or blocklisting. A spam placement tells you the problem is downstream, usually content, engagement history, or sender reputation signals that built up over time.

If you're seeing one and not the other, that's actually useful information. Bounces point to MTA-level rejection. Delivered-but-missing points to inbox-level filtering. Both are solvable, but you need to know which room the problem is in before you start looking for the exit.

Not sure which one you're dealing with? Try our free Email Header Analyzer to see exactly what happened during delivery, or drop us a line at the SOS hotline if something's broken and you need a second set of eyes.

Contributors

Who worked on this answer

Every name links to their profile. Every company links to their site. Real people, real accountability.

Ask an AI · tailored to your setup

Diagnose my filtering problem

My email deliverability is off and I'm not sure if the problem is happening at the MTA level or the inbox level. Based on my situation, can you help me figure out which stage is failing and what signals I should look at to diagnose it? Here's what I'm seeing: [describe your bounce rates, spam placement rates, authentication setup, and any error codes you've received]

Edit the yellow boxes, then send to the AI of your choice.