BreachDetect
← All posts

11 min read

Dark web monitoring for MSPs: how to check client domains

A practical guide for MSPs: get written permission, check client domains for leaked logins, fix urgent hits the same day and turn it into a monthly service.

By BreachDetect Research

On this page
  1. Why MSPs should check client domains for leaked logins
  2. Get written permission first
  3. Use a free check to start the conversation
  4. What to check first across all your clients
  5. The same-day fix in Microsoft 365
  6. What to tell the client
  7. Turn it into a regular service
  8. Common questions
  9. Check a client's domain now

Dark web monitoring for an MSP means searching stealer logs and breach data for leaked logins on your clients' domains, with each client's written permission. You fix the urgent hits the same day and get alerted when new ones appear.

It matters because a client's login rarely leaks from a machine you manage. It comes off a home laptop where someone checked work email, a contractor's PC, or a personal Google account whose saved passwords sync to an infected computer. The client finds out when someone uses it, and they'll expect you to have known first.

The short version:

  1. Get written permission from each client, naming every domain you'll check.
  2. Check your own domain first, then each client's email domains and sign-in pages.
  3. Deal with recent stealer-log logins for the client's own systems the same day.
  4. Revoke sessions before you reset passwords, and deal with the infected device.
  5. Monitor continuously, and send each client a short report every month.

This guide covers each step, including a permission paragraph you can adapt and the Microsoft 365 steps for a same-day fix.

Why MSPs should check client domains for leaked logins

Stealer logs aren't the same as an old breach dump. Malware on an infected computer copies every password saved in the browser, along with session cookies and details about the machine. If anyone ever saved a login to a client's Microsoft 365, VPN or single sign-on page on that computer, it's now in a file someone can buy. Our explainer on what stealer logs are covers how they're made and sold.

Three things make this an MSP problem, not only the client's.

MFA doesn't fully cover it. A password on its own may bounce off a second factor. The session cookie that kept the person signed in often doesn't, which Microsoft describes in its guide to cloud token theft. If you've sold a client on MFA as the answer to stolen passwords, a stealer log is the case it doesn't handle.

The infected machines are usually ones you can't see. In our check of 99 software companies, law firms and hospital systems, 38 had work addresses saved on gaming sites, which points to family computers where someone also checked work email. 56 had logins saved by Android apps. Your endpoint protection never ran on those devices.

It's already there. Every one of those 99 organisations had employee logins posted in the last 90 days, and 75 had a leaked login for their own sign-on systems. They were large, well-known organisations, many with bigger security teams than a typical client has staff.

Get written permission first

Before you look at full results for a client's domain, get their written permission. The results are passwords and personal data belonging to their staff. Our terms require it too: client domains are allowed on the Business and Enterprise plans, and only where the client has authorised you in writing.

A line in your MSA about "security monitoring" is usually too vague. The permission should name the domains, say what you'll look at, say what you're allowed to do when you find something urgent, and say you won't use the findings for anything else.

Here's a paragraph you can adapt. It isn't legal advice, so have your own counsel review it before you use it.

[Client] authorises [MSP] to search stealer logs, breach data and similar sources for leaked logins connected to the domains listed below, including their subdomains. This covers logins that use these domains as email addresses, and logins for websites and apps that [Client] runs or uses under its own account, such as its single sign-on, VPN and webmail pages.

Domains: [list every domain]

[MSP] may view the full records, including passwords, and may use them only to find exposed accounts and devices, secure them, and report to [Client]. [MSP] will not use any credential it finds to sign in to any account, will not share the records with anyone other than [Client], and will delete passwords once the work they relate to is finished.

Where a finding gives access to [Client]'s systems, [MSP] may revoke sessions and reset passwords for the affected accounts straight away, and will tell [named contact] the same day.

This authorisation starts on [date] and lasts until [Client] withdraws it in writing or the services agreement ends, whichever comes first.

Signed for [Client]: [name, job title, date]

A few things make this work in practice:

  • List every domain. Include old domains from acquisitions and rebrands, and any domain staff still send email from. A forgotten domain with working mailboxes is exactly where a leak goes unnoticed.
  • Get it signed by someone with authority. That means an owner, director or the person responsible for IT, not only your day-to-day contact.
  • Agree on the pre-approved actions. Without the paragraph about revoking sessions straight away, you may find a live admin login at 6pm on a Friday and have to wait for a reply before you can act.
  • Fit it with your data processing terms. If you or the client are in the UK or EU, you're handling personal data on the client's behalf, so the permission should sit alongside your data processing agreement.
  • Keep a copy with the engagement file. You'll need to show it before full results for a client domain are released.

Use a free check to start the conversation

You don't need permission to run a free, redacted check. Anyone can check any company domain without an account. The report shows how many leaked logins use the domain, which of the company's own sign-in pages they were for, how many came from stealer logs, and how recent they are. It includes a small sample with email addresses partly masked, and it never shows passwords.

That makes it a fair way to open the subject with a client or prospect. A few suggestions:

  • Bring it to the person who can act on it, not to a shared inbox or the whole team.
  • Lead with the logins for their own systems, listed by hostname. Those are what make it urgent. A large total on its own isn't news: any company with a few hundred staff will have some.
  • Use it to ask for permission, not as a substitute. The free check tells you whether there's a problem. You need the full records, and the client's sign-off, to fix it.

What to check first across all your clients

Start with your own domain

Your technicians' logins are the most valuable ones you look after. A single infected laptop at an MSP can expose your RMM, PSA, remote access and password manager logins, and through them every client you manage. Check your own domain, and the web addresses of your RMM and PSA tools, before anyone else's.

Then each client's domains and sign-in pages

Search each client's email domains and their own websites. A contractor signing in to a client's VPN with a personal Gmail address won't show up in a search for @client.com. It only shows up when you search the VPN's address. Do the same for the client's own Okta address or other company-specific sign-in pages.

Sort the results by urgency

A first search across a dozen clients can return thousands of records. Most of them aren't emergencies. Sort them like this:

WhenWhatWhy
Same dayStealer-log logins for the client's own systems (Microsoft 365, single sign-on, VPN, remote desktop, webmail, admin panels) posted in the last few monthsLikely to work, and the cookies may skip MFA
Same dayAny hit for an admin, finance or other privileged account, or for your own RMM, PSA or remote access toolsOne account reaches a lot
This weekOther stealer-log hits on work addresses, even for third-party sitesThe person's computer was infected, whatever the site
Monthly reportOlder breach-dump hits, accounts that are already disabled, and passwords changed since the record was postedHousekeeping, not an incident

Check whether the password has changed since

You'll want to know whether a leaked password still works. Don't try it. Signing in with someone else's leaked password breaks our terms for accounts you don't administer, and even for ones you do, it can trip alerts and muddy the sign-in logs you're about to investigate.

Compare the date the record was posted with the date the account's password last changed instead. In Microsoft 365, that's the user's lastPasswordChangeDateTime in Microsoft Graph. Okta records it as passwordChanged on the user. If the password changed after the record was posted, the leaked one is probably dead. Treat the device as infected until you know otherwise, because it can leak the new password too.

Look for the same person at other clients

Contractors, bookkeepers and part-time IT staff often work for several of your clients. If one person's machine is infected, search for their email addresses, usernames and IP address across your other clients too. Their logins for those clients may be in the same log.

The same-day fix in Microsoft 365

For any current login to a client's own systems, act the same day. Our playbook for the first 24 hours after a staff login shows up in a stealer log covers this in full. Here's the order for a typical Microsoft 365 client.

  1. Revoke sessions first. In the Microsoft Entra admin center, open the user and choose Revoke sessions, or run Revoke-MgUserSignInSession in PowerShell. A password reset on its own doesn't sign out someone holding a stolen cookie. Microsoft's guidance on revoking access notes that access tokens already issued last an hour by default, so do this before anything else. For Okta, clear the user's sessions. For Google Workspace, sign the user out from the Admin console.
  2. Reset the password, then reset any other company logins saved in the same log, such as the VPN or a shared admin account.
  3. Check the MFA methods. Remove any phone number, authenticator app or device you don't recognise.
  4. Look for signs it's been used, going back to at least the date the record was posted. Check the Entra sign-in logs for new countries, IP addresses and devices. Check mailbox rules with Get-InboxRule, along with new app consents and forwarding. If anything looks wrong, treat it as an incident, not a ticket.
  5. Deal with the device. A company laptop should be isolated through your EDR and rebuilt. If it's a personal or contractor device, tell the owner plainly that it's infected and shouldn't be used for work until it's wiped. Then stop it happening again. A Conditional Access policy that requires a compliant device keeps unmanaged computers out of Microsoft 365. It needs Entra ID P1, which is included in Microsoft 365 Business Premium.

What to tell the client

Tell the client what you found and what you did, in writing, the same day for anything urgent. Keep it short:

  • which domain and which accounts
  • whether the record came from a stealer log or a breach, and when it was posted
  • what you've already done: sessions revoked, passwords reset, device isolated
  • whether you found any sign it had been used
  • what you still need from them, such as who owns the personal laptop involved

Keep passwords out of the email and out of your PSA tickets. Say that the password was exposed and has been reset. A ticket system that every technician can read is the wrong place for a client's leaked passwords.

Turn it into a regular service

A one-off check goes stale quickly. New stealer logs and breach dumps show up every day, and a domain that looked clean in spring can have fresh results by autumn. The useful offer is ongoing monitoring, not a single report.

  • Start with a baseline. The first check on a new client is the biggest. Fix the urgent items, then agree with the client to close older ones in bulk with a reminder about password reuse.
  • Monitor continuously. Add every authorised domain to a watch list and get alerted when new leaked logins appear. Route alerts into your PSA so an urgent hit becomes a ticket, not an email someone misses.
  • Report monthly. Send each client a one-page summary: new records since the last report, how many were for their own systems, which accounts you secured, which devices you dealt with, and anything still open.
  • Recheck when something changes. A suspicious sign-in, a phishing campaign, a new contractor or an acquired company are all reasons to search again.
  • Keep the permission current. When a client adds a domain, get that in writing before you pull full results for it. When a client leaves, stop monitoring their domains and delete any records you exported.

If you want one place to watch every client's domains, that's what we built BreachDetect for. It searches 44 billion stealer-log records from 37 million infected machines, and more than 16 billion records from over 800 breached databases, with new records added every day. The Business plan is the one that allows client domains. It covers 50 domains for $699 a month, or $6,990 a year, with unlimited seats for your technicians and webhooks to send alerts into your own tools.

Common questions

Do I need my client's permission to check their domain?

Not for a free, redacted check, which anyone can run on any company domain. You do need written permission before you see the full records, because they include the client's staff passwords. Our terms require it for client domains.

Is dark web monitoring worth it if the client already has MFA?

Yes. Stealer logs usually include session cookies, and a stolen cookie can let an attacker into an account that's already signed in, without the password or the MFA prompt. MFA makes a leaked password less useful. It doesn't make a stealer log harmless.

How often should an MSP check client domains?

Continuously, with alerts for new results, and a written summary for each client every month. All 99 organisations in our research had something posted in the last 90 days, so a quarterly or yearly check misses most of it.

Can I test whether a leaked password still works?

No. Compare the posting date with the account's last password change, and reset the password if there's any doubt. Signing in with a leaked password isn't allowed under our terms for accounts you don't administer, and it isn't worth the confusion in the logs for accounts you do.

Check a client's domain now

To see what a client's exposure looks like before you ask for permission, run a free check on their domain. It takes a few seconds, you don't need an account, and it never shows passwords.

See what has already leaked from your company.

Search your domain across stealer logs and breached databases. Free, no account needed.