LTVplus

Patch Management Policy for MSPs: What to Automate and What Still Needs a Technician

Image showing different processes and policies

Key takeaways

  • A patch management policy sets the rules for patching: which systems are covered, how fast patches go out, who approves them, and how failures and exceptions are handled.
  • Your RMM can automate scanning, routine approvals, scheduled deployment, and reporting. Failed patches, offline agents, server reboots, and exceptions still need a technician.
  • LTVplus offers Managed Technical Support for MSPs. Dedicated patch, RMM, and backup engineers fix failed and pending patches and keep per-client patch records ready for QBRs and cyber insurance questionnaires.

Say your patch management policy promises critical patches within seven days, and your RMM shows a client at 96% compliance. That looks fine until someone asks about the 14 devices that have been "pending" since last month.

A policy only sets the standard. But your team still has to identify the right patches, deploy them, confirm they installed, and chase the devices that didn't.

This guide gives you a patch management policy template you can adapt for each client. It also walks through the patch process step by step and shows which tasks your RMM can handle and which ones still need a technician.

LTVplus is the managed technical support partner MSPs trust to handle their most important clients. Our patch management engineers take on the failures and exceptions automation leaves behind. For the broader monitoring side of the job, see our guide to NOC-as-a-Service.

What is a patch management policy?

A patch management policy is a documented set of rules for how your MSP identifies, prioritizes, tests, deploys, and verifies software updates.

It includes information on which devices must be patched, how quickly updates must be installed, who approves deployments, and what happens when a patch fails or can't be applied. That covers more than updates. A good policy also accounts for third-party applications, servers, and network devices such as firewalls.

The policy and the process are often used interchangeably, but they do different jobs:

Patch management policyPatch management process
What it definesRules, responsibilities, deadlinesThe steps technicians and tools perform
Main questionWhat must happen, when, and who decides?How does your team carry it out?
ExampleProduction server reboots need the client's approvalSchedule the reboot in the approved window, then confirm services came back
How often it changesWhen risks, contracts, or client requirements changeEvery patch cycle

NIST SP 800-40 Rev. 4 frames patching as preventive maintenance, which is a normal part of running technology rather than an emergency task. MSPs have it harder because you're running one policy framework across many clients. Each one has its own maintenance windows, business applications, and approvers.

Patch management policy template for MSPs

Copy this structure into your PSA or documentation platform, then create a version for each client. Replace the bracketed placeholders with that client's requirements.

Policy sectionWhat you need to defineExample
1. Purpose and scopeClients, sites, operating systems, applications, and network equipment covered, plus any exclusionsAll managed Windows and macOS endpoints, Windows and Linux servers, approved third-party apps, and firewalls in [Client]'s asset inventory
2. Asset inventoryWhere the inventory lives, who maintains it, and how new or retired devices are recordedReconcile the RMM device list against the documentation platform monthly. Investigate any device that appears in only one.
3. Roles and responsibilitiesWho deploys, tests, verifies, and reports, and which decisions need client sign-offMSP technician deploys and verifies workstation updates. Client IT director approves production server reboots
4. Risk tiers and timelinesRemediation deadlines based on exploitation, exposure, and business impactTier 1: [X] days. Tier 2: [X] days. Tier 3: [X] days. Emergency vulnerabilities follow an expedited process
5. Maintenance windowsDeployment and reboot windows per client and device group, plus emergency proceduresWorkstations: [window]. Production servers: [client-approved window]. Emergencies escalate to [contact]
6. Testing and deployment ringsPilot group, broader groups, and when updates move between themPilot on IT staff devices. Broad rollout after [X] days with no significant issues
7. Approvals and change controlWhat auto-approves, what needs technician review, and what needs client authorizationRoutine workstation security updates auto-approve. Server, firmware, and business-critical app updates need documented approval.
8. Exceptions and compensating controlsHow deferrals are requested and approved, what control covers the gap, and when it expiresEach exception is logged in the PSA with a reason, owner, compensating control, and expiry date
9. Failures and rollbackRetry limits, investigation, escalation path, and rollback criteriaTwo failed attempts open a PSA ticket. Unresolved critical failures escalate to [engineer] within [X] hours
10. Reporting and evidenceReport frequency, what's measured, and how long records are keptMonthly compliance report covering failures, pending reboots, and open exceptions. Quarterly trend in the QBR
11. Review cadenceHow often the policy is reviewed, who approves changes, and what triggers an extra reviewQuarterly, and every after a major incident, infrastructure change, or new contractual requirement

How to set the timelines in Section 4

If you need a reference point for risk tiers, look at CISA's Binding Operational Directive 26-04 (BOD 26-04), issued in June 2026. It replaced severity-score deadlines with four questions:

  1. Is the vulnerable asset publicly exposed?
  2. Is the vulnerability in CISA's Known Exploited Vulnerabilities catalog?
  3. Can an attacker automate every step of the exploit?
  4. Would exploitation give an attacker partial or total control of the asset?

CISA uses the combination of answers to set deadlines, from three days for the highest-risk flaws to the asset's next scheduled major upgrade for the lowest. The directive only binds U.S. federal civilian agencies, but it gives you a published model to point to when you agree on timelines with each client.

Step by step patch management process

Run these nine steps every patch cycle, for every client.

Step 1: Inventory. Confirm every managed device is in the RMM, assigned to the right client, and reporting in. Flag anything missing or inactive.

Step 2: Scan. Run scheduled scans for missing OS and third-party updates, plus extra scans after onboarding new devices or restoring from backup.

Step 3: Assess and prioritize. Sort updates by your risk tiers: KEV status, exposure, and how important the device is to the client.

Step 4: Test. Release updates to the pilot ring first to catch failed installs, driver conflicts, and app issues.

Step 5: Deploy. Push approved updates to the remaining rings inside each client's maintenance window. Anything outside the window follows the client's emergency change procedure.

Step 6: Verify. Check install records and reboot status. For servers, also confirm that key services and applications are running.

Step 7: Remediate failures. Investigate everything marked failed, pending, or incomplete, and record the fix in your PSA so recurring problems surface.

Step 8: Report. Show which devices are compliant and separate failures, pending reboots, offline devices, and approved exceptions.

Step 9: Review exceptions. Before closing the cycle, check that every open exception still has an owner, a reason, and a valid expiry date.

What can you automate in patch management?

Most routine patching follows predictable rules, which makes it a strong fit for automation. Once you set those rules, your RMM can run most of Steps 2, 4, 5 and 8 on its own. RMM platforms such as NinjaOne, N-able, ConnectWise, Atera and Kaseya do it through:

  • Recurring schedules that check every endpoint for updates without anyone starting the job
  • Approval policies that release routine security updates automatically and hold riskier ones until a technician signs off
  • Device groups that move an update from your pilot devices to everyone else once the waiting period passes
  • Built-in software catalogs that keep supported third-party apps current
  • Agent alerts that open a ticket or notify your team when an endpoint goes quiet
  • Report templates that build and send each client's patch summary on a set date

The rest of the cycle (especially fixing failures and reviewing exceptions) still needs your team.

If you're comparing platforms, our breakdown of PSA and RMM tools covers the main options. For scripting routine fixes, see our guide to Kaseya VSA scripting and automation.

What still needs a technician?

An RMM can detect a failed patch, retry it, and send an alert. Some platforms can also run remediation scripts for errors your team has seen before. That handles the problems you can predict.

However, patch failures are not that predictable. The same error code can come from a full disk on one device and a broken driver on another, and a script written for one cause won't fix the other. Your RMM also doesn't know which client runs a fragile accounting app, which server can't go down during month-end close, or who has to approve a reboot. That context lives with your team and in each client's policy.

So technician work splits into two groups: problems the RMM flags but can't finish, and decisions that depend on the client's business and risk.

Work the RMM flags but can't finish:

  • Failed or pending patches. Updates fail because of low disk space, corrupted update files, driver conflicts, or missing prerequisites. Retrying rarely fixes the cause. The technician has to find it, whether that's a full disk or a broken driver, then confirm the update actually installed.
  • Offline, stale, or missing agents. No agent means no patching and no current status. A technician has to find the device and get the agent reporting again.
  • Server reboots. Your RMM can schedule a reboot, but production servers need a technician to follow the client's change process and confirm that services and applications came back.

Decisions that need judgment:

  • Business application testing. An OS update can install cleanly and still break an app the client depends on. Your team needs to know what runs on the system and test with the client before a wider rollout.
  • Firewall and edge-device firmware. Update methods vary by vendor and often sit outside your workstation patch policies. Each update needs a check against the vendor's advisory and a plan for any connectivity disruption.
  • Rollback decisions. Some patches can't be cleanly reversed, and removing a security fix reopens the vulnerability. Choosing between uninstalling, restoring, or applying a workaround takes judgment.
  • Out-of-band emergency patches. Actively exploited flaws can't always wait for the next window. A technician decides whether to deploy now or apply a temporary mitigation, with the client's approval.
  • Exceptions. When a client needs to defer a patch, the risk has to be documented and covered by a compensating control.

Do your patch reports keep showing the same pending and failed devices? LTVplus patch and RMM engineers work inside your tools to clear failed and pending patches, under your brand. See LTVplus patch management services.

Patch management best practices for MSPs

Prioritize by exploitation, not just severity score

  • A CVSS severity score tells you how bad a flaw could be. It doesn't tell you whether attackers are using it. The CISA KEV catalog lists vulnerabilities with evidence of exploitation in the wild, so anything on it that affects a client should go on your fastest track.
  • Verizon's 2026 Data Breach Investigations Report found that exploitation of vulnerabilities is now the most common way attackers get in.

The four BOD 26-04 questions under the template give you a simple way to apply this across clients.

Keep edge devices on their own track

  • Firewalls, VPN appliances, routers, and switches control how traffic gets in and out of a client's network, so an unpatched one raises the client's exposure. Don't assume your workstation patch policy covers them.
  • Give edge devices a separate schedule and a named owner for each device group. Their section of your vulnerability and patch management policy should cover the inventory, vendor advisory monitoring, configuration backups before updates, and post-update checks.

Map your policy to the frameworks clients ask about

Clients in regulated industries often need to show how they find and fix vulnerabilities. Two references come up most:

LTVplus is ISO 27001:2022 certified. That certification covers our own security practices, not the compliance of the MSPs or clients we support.

Try this: Pull your RMM's list of patches that have been pending for more than one maintenance cycle and sort it by client. The clients at the top are where your written policy and your day-to-day patching have drifted furthest apart.

Use patch reports to show clients the work

Patch reports show work your clients never see. A single compliance percentage is a snapshot, so in your QBR, show the following:

  • Compliance by device group
  • Critical patches still outstanding
  • Failures and how they were resolved
  • Open exceptions and deferred patches
  • Devices that stopped reporting

If compliance dropped, explain why: new devices, failed deployments or updates deferred for compatibility. If your contracts set patching deadlines, these reports also help you prove SLA compliance.

The same records help clients at cyber insurance renewal, since applications often ask about patching practices. Keep each client's approved policy, recent patch reports and exception log ready.

Pro tip: Keep a per-client evidence folder and update it after every patch cycle. When renewal comes around, you're handing over records you already maintain instead of rebuilding months of activity from old RMM data.

Hand off the patching work your RMM can't finish

Your patch policy only works if someone owns the failures, offline agents, reboots, and exceptions every week.

LTVplus gives your MSP dedicated patch, RMM, and backup engineers who work inside your existing PSA, RMM, and documentation tools, under your brand. They set and maintain patch policies for each client, investigate failed and pending patches, restore offline or missing agents, and patch servers inside your approved maintenance windows. Reboots follow your change process. You keep the client relationship and the escalation decisions.

Dedicated engineer seats start at $3,000 per month, with business-hours, extended, and 24/7 coverage. You can start with a 3-month pilot to see how it fits your operation. See LTVplus pricing.

Book a scoping call.

Frequently asked questions

What should a patch management policy include?

A patch management policy should include scope, asset inventory, roles, risk-based timelines, maintenance windows, testing rings, approvals, exceptions, failure handling, reporting, and a review schedule.

How often should MSPs patch client systems?

Microsoft releases Windows security updates on Patch Tuesday, the second Tuesday of each month, so that's a natural anchor for a regular cycle. Set each client's deployment window around its risk and change process. Handle critical and KEV-listed vulnerabilities through a separate emergency process.

What is the difference between a patch management policy and a process?

The policy sets the rules: what's covered, who approves, and how fast patches must go out. The process is the set of steps your team runs every cycle to meet those rules.

Can patch management be fully automated?

When it comes to patch management, most routine work can be automated. However, the judgment calls, like rollbacks and exceptions, still need a technician.