In a Microsoft 365 incident, end the attacker’s access first.
- We revoke sessions, reset credentials and remove what the attacker added: inbox rules, forwarding, sign in methods and app consents.
- Then we read your sign in and audit logs to find what was opened, sent and changed.
- You get a hardened tenant and a written record for leadership, your insurer or counsel.
After an incident, or preparing for one? Book a 30 minute call
200+ Microsoft 365 and Azure tenants. 100+ engagements. 20+ years in regulated technology.
How it usually starts
Four common first signs, and what each one tends to mean.
“I signed in from the link in that email. It asked for my code, like always.”
That is how adversary in the middle (AiTM) phishing works. A proxy page passes the real sign in through, MFA included, and keeps the session it produces.
“The vendor says they never sent new bank details.”
Payment fraud attempts can run from inside a real mailbox, with inbox rules that hide the replies from its owner. The first questions are which mailbox, and since when.
“Someone approved an app none of us recognize.”
An app a user consented to keeps its access after a password reset and after MFA, so the grant itself has to be found and removed.
“Defender raised an alert on one of our accounts.”
An alert is a starting point, not a verdict. The sign in and audit logs show whether it was real and how far it went.
Read more: how AiTM token theft works in Microsoft 365
Not sure it is a compromise? Run the account takeover check, or call (786) 577-9743 now.
What happens when you call
Containment first, then scoping, then recovery. How long each part takes depends on what we find, so each step says what sets the pace instead of a length.
-
Contain
What sets the pace How many accounts are involved, whether an admin account is one of them, and whether accounts sync from on premises Active Directory
We end the attacker’s access before we study it. Sessions and refresh tokens are revoked, the password is reset (twice for accounts synced from on premises Active Directory), and sign in methods and devices the user did not add are removed.
Then what they left behind: inbox rules, including hidden ones, mailbox forwarding, and app consents granted during the attack. The addresses they signed in from are blocked with Conditional Access. Each item is exported before it is removed, so the evidence survives. If disabling an account would stop the business, we agree the order with you first.
You get The attacker’s access ended, and a log of every change we made
-
Scope
What sets the pace How long the attacker had access, how many mailboxes and sites they reached, and what your licenses and log settings kept
We use the Session ID in the Entra sign in logs to separate the attacker’s activity from the user’s, then follow it into the Purview audit log: mail read and sent, inbox rule changes, files opened in SharePoint and OneDrive, and changes made in Entra ID.
We also check whether the same email, address or app reached anyone else. If a payment thread was touched, the other party should hear about it by phone, not by email.
You get A timeline of how they got in, and what they read, sent and changed
-
Recover and harden
What sets the pace Your licenses, how your people sign in today, and how many devices need to be enrolled first
We close the gap that let the session be replayed: phishing resistant MFA, such as passkeys or Windows Hello for Business, required through Conditional Access, admins first, and a compliant device where it fits. Emergency access accounts stay out of blocking policies.
Token protection, which needs Entra ID P1, binds sessions to a registered device for supported desktop and mobile apps. It does not cover browser sessions to Outlook on the web or SharePoint, so it is one layer, not the fix.
Then alerts in the Microsoft Defender portal, and in Microsoft Sentinel if you use it, for new forwarding, inbox rules, app consents and sign in methods, sent to someone who reads them. What is available depends on your licenses.
You get Controls that close the way in, tested in report only mode before they are turned on
-
The written record
What sets the pace Who will read it: leadership, your insurer, your counsel, or all three
A plain account of what happened, what was accessed, what we changed and what is still open, with the timeline and the exports behind it. Whether anyone has to be notified is a decision for you and your counsel, and the record gives them the facts to make it.
You get The incident record, the evidence exports and a list of open items
What we need from your side
An admin who can grant us the roles the work needs, a person who can approve disabling accounts, your license list, and someone who knows which payments or contracts were in flight. If accounts sync from on premises Active Directory, we also need someone with access there.
Forensic imaging of laptops and servers, and legal advice, stay with specialists. When your insurer or counsel brings in a forensics firm or a breach coach, we work alongside them. Your point of contact is a senior security and compliance lead who also directs the engineering in Microsoft 365 and Azure.
Read more: passkeys and the SMS MFA retirement, Microsoft 365 evidence for cyber insurance
Work we have delivered
Past incident and access work, without client names.
-
Incident response
A full Microsoft 365 incident response after an adversary in the middle (AiTM) session token theft.
-
Privileged access
Standing admin risk that became a funded Privileged Identity Management (PIM) program.
With PIM, admin roles are activated when needed, for a set time, instead of held all the time. It needs Entra ID P2 or Entra ID Governance.
What the written record looks like
Three parts of an incident record. Each one is a sample format with illustrative content, not a client’s document.
Incident timeline
Each event, where it was found and what was done about it. Times are in UTC so every log lines up.
Incident timeline, one mailbox
Sample format, illustrative content
| Time, UTC | What happened | Found in | Action |
|---|---|---|---|
| 14:09 | A user signs in through a link in an email. MFA completed. | Entra sign in logs | Account disabled, sessions revoked |
| 14:11 | The same Session ID is used from a hosting provider address | Entra sign in logs | Address blocked in Conditional Access |
| 14:20 | An authenticator app the user did not add is registered | Entra audit logs | Method removed, password reset |
| 14:26 | An inbox rule moves mail from a supplier to RSS Subscriptions | Purview audit log | Rule exported, then removed |
| 14:40 | An open invoice thread is accessed | MailItemsAccessed records | Supplier called to confirm bank details |
Containment checklist
What was checked, what was found and what is still open.
Containment checklist
Sample format, illustrative content
- Sessions
- Done Revoked, account disabled during cleanup
- Password
- Done Reset twice, account synced from on premises
- Sign in methods
- Done One authenticator app removed
- Rules, forwarding
- Done One rule removed, no forwarding found
- App consents
- Done None granted during the attack
- Other mailboxes
- In progress Same email found in three more inboxes
Summary for leadership
One page in plain language, with the decisions that belong to you.
Incident summary
Sample format, illustrative content
- What happened
- A phishing page relayed one finance user’s sign in and kept the session.
- What was accessed
- Mail in that account, including one open invoice thread.
- What we changed
- Account secured, rule removed, address blocked, phishing resistant MFA required for admins.
- Still open
- Passkeys for the finance team. Device compliance policy in report only mode.
- Your decisions
- Open Whether to notify anyone, with your counsel and insurer
No active incident? Prepare for one
Three things worth doing before you need them, all done with your IT team or MSP over calls.
A tabletop exercise
On a call, we walk the people who would make the decisions through one realistic scenario, such as an AiTM phishing email reaching accounts payable. Who disables the account, who calls the bank, who tells leadership and the insurer.
You get Gaps found in an exercise, not during an incident
A response plan for Microsoft 365
Who can disable accounts and revoke sessions, where the emergency access accounts are and who holds them, which numbers to call first, and the order of the first steps. Short enough to use under pressure.
You get A plan your IT team or MSP can follow
Logging and retention checks
We confirm audit logging and mailbox auditing are on, check which mailbox actions your licenses record, and see how long each log is kept. If 30 days of sign in logs is too short, they can be sent to Microsoft Sentinel or a Log Analytics workspace.
You get Logs that still exist when you need them
Last checked
How long Microsoft 365 keeps the logs an investigation needs
Upgrading a license later does not bring back logs that have already expired.
| Log | What it shows | Kept for |
|---|---|---|
| Entra sign in logsMicrosoft Entra admin center | Where and how each sign in happened, with the Session ID that ties a stolen session together | 7 days without Entra ID P1 or P2. 30 days with P1 or P2, and Business Premium includes P1. |
| Entra audit logsMicrosoft Entra admin center | New sign in methods, app consents and role changes | 7 days without P1 or P2. 30 days with P1 or P2. |
| Audit (Standard)Microsoft Purview | Mailbox actions, inbox rule changes, and file activity in SharePoint and OneDrive | 180 days |
| Audit (Premium)Microsoft Purview, with E5, Purview Suite or the E5 eDiscovery and Audit add on | The same records, kept longer | One year for Entra ID, Exchange, OneDrive and SharePoint records. Up to 10 years with an extra add on license. |
| Microsoft Sentinel or Log AnalyticsAzure, sent through diagnostic settings | Entra sign in and audit logs, kept past 30 days | As long as the retention you set, with Azure costs |
Want the rest of the tenant checked too? Microsoft 365 security assessment
Questions people ask first
We think an account is compromised right now. What should we do?
Call (786) 577-9743 rather than booking a time. If you can, turn off sign in for the account in the Microsoft Entra admin center and select Revoke sessions while we talk.
Is resetting the password enough?
No, because inbox rules, forwarding, app consents and sign in methods the attacker added all survive a password reset. Containment removes each of them and revokes the account’s sessions as well.
Can you tell us what the attacker read?
That depends on what your logs kept. The Entra sign in logs and the Purview audit log, including MailItemsAccessed records where your licenses capture them, show what the attacker’s session opened, sent and changed.
Should we call our cyber insurer first?
Read your policy, and if it asks you to notify the insurer or to use firms it chooses, do that as well as calling us. The written record we keep is built so your insurer and counsel can follow it.
Do you replace our IT team or MSP?
No. We work beside them during the incident and hand back a record of every change we made.
Can you guarantee the attacker is gone?
No one can honestly promise that. We remove every foothold we find, record it, and set up alerts for the signs of a return.
On the first call
- Tell us what you have seen and which accounts are involved.
- We agree admin access and who can approve changes.
- Containment starts with the account most at risk.
If it is happening now, call rather than book.
We start with containment, beside your IT team or MSP. For a past incident, or to prepare for the next one, book a 30 minute call.