Reducing Phishing Risk in Enterprise Email

Which controls reduce phishing risk? SPF, DKIM and DMARC, inbound mail filtering, MFA and session security, user reporting and call-back verification for IBAN changes.
Technical summary
Technology
SPF, DKIM, DMARC, Defender for Office 365
Topic
Email security
Level
Start
In this article6 sections

Phishing remains one of the most common starting points for attacks on organizations. The reason is simple: email reaches every employee, and a single click is enough. No single product closes this risk. Unless domain authentication, technical filtering, account security, user behavior and business processes are addressed together, one link in the chain stays open.

1. Protect your domain: SPF, DKIM, DMARC

Record Role
SPF Specifies which servers can send email on behalf of your domain
DKIM Signs email with a key belonging to your domain
DMARC Specifies what the recipient should do when SPF/DKIM alignment fails and where reports should go

Start DMARC with p=none and review the reports; after aligning all legitimate sending services (CRM, newsletter, website forms), move to quarantine and reject policies. Since 2024, Google and Yahoo have required these authentications from bulk senders; incomplete configuration can cause your own emails to be rejected too.

A typical starting DMARC record looks like this: v=DMARC1; p=none; rua=mailto:[email protected]. In an organization using Microsoft 365, the simplest SPF record is v=spf1 include:spf.protection.outlook.com -all; other sending services are added to this record.

Common mistakes on the DNS side

  • Publishing two separate SPF records for the same domain. SPF evaluation errors out and the result may be as if there were no record at all.
  • Exceeding the 10 DNS lookup limit in SPF. Every include chain counts; when the limit is exceeded, validation fails.
  • Enabling DKIM only for the main mail system and forgetting third-party services that send invoices and newsletters.
  • Leaving DMARC at p=none for years. Reports get collected, nobody reads them, and your domain stays open to spoofing.
  • Writing p=reject directly without reviewing reports. On the first day, quote emails sent from the CRM disappear.

Protect not only the domains you send email from but also the domains that never send email. A domain bought for a campaign or used for redirects can easily be used for spoofed sending if it has no SPF and DMARC records. For these domains, v=spf1 -all and a p=reject policy are sufficient.

2. Filter inbound mail

  • Links checked at the time of click and attachments analyzed in a sandbox.
  • Impersonation protection for executives and the finance team.
  • A visible warning tag on external emails.
  • Rules for mail coming from lookalike domains (e.g. swapped letters).

The filter’s worst enemy is exception lists. Adding a whole domain or IP block to the allowlist in response to the complaint “this vendor’s mail keeps going to quarantine” carries the attacker’s mail straight to the inbox once that vendor’s account is compromised. Review the allowlist periodically and prefer individual rules to broad exceptions.

On the attachment side, Office files with macros, password-protected archives and HTML attachments should be handled separately. A password-protected archive can’t go through the sandbox. Question whether such attachments are a business need; if not, quarantine them.

3. Make account takeover harder

The target of phishing is often the password and session information. MFA, disabling legacy protocols and Conditional Access policies ensure that a compromised password alone isn’t enough. For stronger protection, phishing-resistant methods such as FIDO2 security keys or Windows Hello for Business can be considered.

Classic MFA doesn’t stop every attack. Adversary-in-the-middle (AiTM) phishing pages sit between the user and the real sign-in page, get the user to complete the MFA approval and steal the resulting session cookie. Neither an SMS code nor a push approval helps in this scenario. FIDO2, however, is bound to the domain; it doesn’t work on a fake page. That’s why it makes sense to move to a phishing-resistant method at least for executives and the finance team.

Another door is OAuth consent attacks. The user doesn’t give away their password; they grant a fake application permission to read their mailbox. Even if the password changes, the permission persists. Restrict users from consenting to unverified applications and tie consent requests to an admin workflow.

4. Make the user part of the system

  • A one-click “report as suspicious” option in Outlook and on mobile.
  • Quick review of reported emails and purging similar ones from all mailboxes.
  • Short, role-based training and authorized phishing simulations; the goal is to raise the reporting rate, not to lower the click rate.

A report button only works if reports get a response. If a user reports a suspicious email and gets no feedback, they won’t report the next one. Define who reviews reported emails and how quickly. Exposing users who click after a simulation also backfires; users learn to hide their mistakes. A single click reported early is far cheaper than an account takeover noticed hours later.

5. A process against invoice fraud (BEC)

Technical controls can’t catch every fake email. IBAN or bank detail change requests should be verified with a call-back to the registered phone number; contact details in the email shouldn’t be used. This single rule prevents a large share of the most expensive losses.

In a BEC attack, the attacker often uses a real supplier’s compromised account. The email passes SPF, DKIM and DMARC, because it really does come from that domain. The correspondence history is real and the invoice format is familiar. A technical filter can’t tell the difference. The process is what will. Rules such as dual signatures on payment approval, an additional check on requests with urgency pressure and manager approval for the first payment to a new IBAN should be put in writing.

When a phishing email is clicked

  1. Reset the user’s password and end all their sessions.
  2. Check whether there are changes to MFA methods and mailbox rules.
  3. Identify the other users who received the same email and purge it.
  4. Review the EDR alerts on the device.

Resetting the password alone isn’t enough; a stolen session cookie stays valid. In Microsoft 365, sessions can be revoked from the Entra admin center or with the Revoke-MgUserSignInSession command in Microsoft Graph PowerShell. A common trace in mailbox rules is rules that delete emails containing certain keywords (invoice, payment, IBAN) or move them to the RSS folder, and forwarding to an external address. With these rules, the attacker delays the user noticing. The attacker having added their own device to the registered MFA methods should also be on the checklist.

For technical configuration, see our Email Security and Phishing Protection page; for the user side, Cybersecurity Training and Phishing Simulation; and for the identity side, Microsoft 365 Security.

Support

Contact us for support or a discovery session on this topic.

Contact form
[email protected]

Eretek Cybersecurity

Firewall, VPN, endpoint security and threat prevention. We reduce the attack surface and measure and report the risks.

Knowledge Center

Other articles