Microsoft 365
How do you set up Microsoft 365 Advanced Delivery for phishing simulations?
Updated
Short answer
In the Microsoft Defender portal, go to Email & collaboration, Policies & rules, Threat policies, Advanced delivery, then the Phishing simulation tab. Add the simulation vendor's sending domain (up to 50) and sending IP addresses (up to 10). Messages that match at least one domain and one IP are delivered without filtering, and Safe Links will not block or detonate their links at time of click.
What does the advanced delivery policy do?
Microsoft 365 is secure by default, which means it does not normally allow allow lists or filtering bypass for messages it identifies as malware or high confidence phishing. Phishing simulations are designed to look like phishing, so without an exception they can be quarantined, flagged or have their links blocked.
The advanced delivery policy is Microsoft's supported exception for two cases: third-party phishing simulations and SecOps mailboxes. For matching simulation messages, Microsoft documents the following results.
- Filters in Microsoft 365 take no action on the messages.
- Zero-hour auto purge (ZAP) for spam and phishing takes no action.
- Safe Links does not block or detonate the URLs at time of click. URLs are still wrapped, but not blocked.
- Safe Attachments does not detonate attachments.
- Default system alerts are not triggered, and automated investigation (AIR) ignores the messages.
- If a user reports the message with the built-in Report button in Outlook, no alert or incident is created, but it appears on the User reported tab of the Submissions page.
Where is Advanced Delivery in the Defender portal?
Open the Microsoft Defender portal at security.microsoft.com and go to Email & collaboration, then Policies & rules, then Threat policies, then Advanced delivery in the Rules section. You can also go straight to security.microsoft.com/advanceddelivery.
The page has two tabs: SecOps mailbox and Phishing simulation. For a third-party simulation platform, use the Phishing simulation tab. If no entries exist yet, select Add. If entries already exist, select Edit instead, because the Add button is not shown.
Which admin roles do you need?
Microsoft lists several ways to get the right permissions. Use the least privileged option that works for your tenant.
- Email & collaboration and Exchange Online permissions: Security Administrator in Email & collaboration RBAC plus Organization Management in Exchange Online RBAC to create, change or remove settings.
- Microsoft Entra roles: Security Administrator gives the required access. Global Administrator also works, but Microsoft recommends limiting that role to emergencies.
- Microsoft Defender XDR Unified RBAC (if active for Defender for Office 365): Authorization and settings/Security settings/Core Security settings (manage).
- Read-only access: Global Reader or Security Reader, or View-Only Organization Management in Exchange Online.
What do you enter for domains, IPs and simulation URLs?
Your simulation vendor should give you these values. If you want to check them yourself, send a test simulation and read the Authentication-Results header: the sender IP, smtp.mailfrom and the DKIM header.d value are the fields you need.
- Domain (required, up to 50 entries): the domain in the 5321.MailFrom address (also called the MAIL FROM, P1 sender or envelope sender), or the DKIM domain specified by the vendor. Use Punycode for internationalized domain names.
- Sending IP (required, up to 10 entries): IPv4 as a single address, a range or CIDR notation. IPv6 addresses are currently supported only through PowerShell.
- Simulation URLs to allow (optional, up to 30 entries): only for non-email simulations, such as links in Teams messages or Office documents. You do not need these for email simulations, because phishing simulation URLs in email are automatically allowed during mail flow and at time of click.
How does matching work, and how do you confirm it?
A message must match at least one Domain and at least one Sending IP. Microsoft notes that no association between values is kept, so any listed domain combined with any listed IP counts as a match.
After saving, select Add or Save, review the confirmation flyout and close it. Then send a test. In Threat Explorer or Real-time detections, filter on System override source and choose Phishing simulation to confirm the message was recognized. The Email entity page also shows Phishing simulation under Tenant override.
In Exchange Online PowerShell, the same setup uses New-PhishSimOverridePolicy followed by New-ExoPhishSimOverrideRule with the Domains and SenderIpRanges parameters. Get-ExoPhishSimOverrideRule shows the current rule.
What does advanced delivery not cover?
Advanced delivery solves most delivery problems, but Microsoft documents several limits that matter for MSPs managing many tenants.
- Malware filtering and ZAP for malware are bypassed for SecOps mailboxes only, not for phishing simulations.
- If your MX record does not point to Microsoft 365, the IP in the Authentication-Results header must match the policy. You may need Enhanced Filtering for Connectors. Do not add your gateway's own IPs as a workaround, because Microsoft warns this effectively bypasses spam filtering for anyone spoofing the simulation domain.
- Enhanced Filtering for Connectors does not work for simulations when mail goes from the internet to Microsoft 365, out to another service and back again.
- Direct Injection emails bypass the transport pipeline, so the policy does not apply to them.
- Simulations within the same organization (DIR:INT) are currently not supported, especially with Exchange Server hybrid routing.
- Do not add simulation URLs to the Safe Links Do not rewrite list. Microsoft says this might cause unwanted URL click alerts.
Does advanced delivery stop false clicks from link scanners?
Not completely. The policy stops Safe Links from blocking or detonating simulation links at time of click, and Microsoft notes URLs are still wrapped. It does not stop every automated system from opening links, and it does nothing about third-party gateways or other tools in the mail path. A simulation platform still needs its own way to tell scanner visits from real people.
NetMinder Security, a white-label phishing simulation platform for MSPs on Microsoft 365, works alongside this policy. It checks every link visit for timing, browser and network signals so that scanner traffic is filtered out of click results.