← All posts
Managed ITEmail security

How Email Filtering Works: A Guide for Business

The checks every message goes through, what they miss, and the settings worth looking at.

Somebody in your office opens Outlook in the morning and sees 40 new messages. What they don’t see is everything that arrived overnight and never made it: the fake parcel notices, the “your mailbox is full” links, the attachment pretending to be an invoice. Those were stopped in the second or two between the message reaching your email service and landing in the inbox.

That’s email filtering. Most businesses have some of it without knowing how it’s set up, which is fine until a real client’s email goes missing or a convincing fake gets through. This guide explains what the filter is doing, so you can ask better questions about yours.

The checks on the way in

Every incoming message goes through a series of checks, roughly in this order. The details below follow how Microsoft 365 does it (Microsoft Learn), since many small businesses use it, but other filtering services work along the same lines.

1. Who is sending it

Before reading the message, the filter looks at the server sending it. Is it on a list of known spam sources? Has it ever sent mail before? Is it suddenly sending thousands of messages an hour? A lot of junk is turned away at this stage.

2. Is the sender who they claim to be

Email was designed with no proof of who sent it, so anyone can put any name in the From line. Three standards fill that gap, and Microsoft describes them as building blocks that work together (Microsoft Learn):

  • SPF (Sender Policy Framework) is a published list of the servers allowed to send mail for a domain.
  • DKIM (DomainKeys Identified Mail) is a digital signature on the message, showing it came from the domain and wasn’t changed on the way.
  • DMARC (Domain-based Message Authentication, Reporting and Conformance) tells receiving servers what to do with mail that fails those checks, and sends the domain owner reports.

These matter in both directions. They help your filter spot fakes, and if your own domain doesn’t have them set up properly, other companies’ filters are more likely to junk your real mail and someone can more easily send email pretending to be you.

3. What’s in the message

Next the filter looks at the content: the wording, the links, the layout, and whether it matches patterns seen in spam and phishing elsewhere. Large services see huge volumes of mail, so a new scam campaign spotted at one company can be blocked for everyone soon after.

Attachments are scanned for known malware. Better filters also open them in an isolated space to see what they do before the message is delivered. Links can be checked against lists of known bad sites, and with some services they’re checked again at the moment someone clicks, which catches a page that turned malicious after the email arrived (Microsoft Learn).

5. A second look after delivery

Some messages look fine when they arrive and are recognized as phishing an hour later. Microsoft 365 can pull those back out of inboxes after delivery, a feature it calls zero-hour auto purge (Microsoft Learn). It’s on by default, and it’s worth leaving on.

Where stopped mail goes

A message that fails these checks usually goes one of three places:

  • The Junk Email folder, for messages that are probably spam. Users can still find them.
  • Quarantine, a holding area outside the mailbox, for messages more likely to be harmful. In Microsoft 365, an administrator decides how long they’re kept (anywhere from 1 to 30 days) and whether users can release their own or must ask.
  • Rejected outright, for messages that are clearly malicious.

The quarantine settings are worth a look. If staff can release anything themselves, a determined person will release the fake invoice. If they can’t release anything, they’ll stop reporting missing mail and start using personal email to get things done.

If your own domain doesn’t have them set up properly, other companies’ filters are more likely to junk your real mail.

Outbound filtering

Filtering also watches what leaves. If a mailbox in your company suddenly starts sending hundreds of messages, that usually means the account has been taken over, and Microsoft 365 will block it from sending and alert an administrator (Microsoft Learn). That’s protection for your reputation as much as anything: if your domain is seen sending spam, other companies start blocking your legitimate email too.

When a staff member says “I can’t send email and I got an odd warning,” don’t just unblock them. Check the account first. Our post on Microsoft 365 account takeover covers what to look for.

False positives: when real mail is stopped

No filter is perfect, and a strict one will sometimes stop mail you wanted. The usual reasons:

  • The sender’s domain isn’t set up properly. A small vendor without SPF or DKIM looks a lot like a fake. The real fix is on their side, and it’s worth telling them.
  • Bulk-looking mail. Newsletters, statements and automated notices can score as bulk mail. Microsoft rates each message on a scale for this, and the threshold can be adjusted.
  • Forwarding and mailing lists that change messages on the way through, which can break the signature checks.

The tempting fix is to add the sender to an allow list. Do that sparingly, and for specific addresses rather than whole domains, because an allow entry also lets through anyone who manages to fake that sender. Microsoft’s own guidance treats allow entries as exceptions to review regularly.

What filtering doesn’t catch

Filtering is one layer, and it has limits worth knowing.

  • Mail from a real account that’s been hijacked. If a vendor’s mailbox is taken over, the fake invoice comes from their genuine address and passes every check. Business email compromise of this kind cost US businesses just over $3 billion in reported losses in 2025, according to the FBI (IC3 2025 report). A phone call to confirm bank changes is the only reliable defense.
  • Clean messages with no link or attachment. “Are you at your desk? I need a quick favor” has nothing for a filter to scan.
  • Pages that turn bad later. Click-time checks help, and so does DNS filtering, which blocks known bad sites whatever route the link took. See our post on DNS filtering.

Questions to ask about your filtering

  1. Are SPF, DKIM and DMARC set up for our domain, and is DMARC set to act on failures, not just report them?
  2. Are links checked when someone clicks, or only when the email arrives?
  3. Who can release mail from quarantine, and who reviews what’s held?
  4. Is post-delivery removal switched on?
  5. What’s on our allow list, and when was it last reviewed?
  6. Will someone be alerted if one of our accounts starts sending spam?

If you’re comparing filtering products rather than tuning the one you have, our post on email spam filtering solutions looks at the options.

How we handle email filtering

For our Managed IT customers, email filtering is one layer of our cybersecurity service: filtering that helps keep phishing and malicious messages away from your team, including after they reach the inbox. We also look after the parts around it that people forget, like your domain’s SPF, DKIM and DMARC records, quarantine settings and outbound alerts, as part of Cloud and Microsoft 365 support. If your real email is landing in other people’s spam, or fakes keep getting through, call us at (386) 252-2292.

More on Managed IT

Let's build your foundation.

Tell us where your business is today and we'll help map the next step.