Switching EDR Vendors Without a Coverage Gap: A 10-Step Migration Playbook
Swapping endpoint platforms can open a gap in protection and in your insurance answers. This 10-step sequence keeps every device covered and every renewal statement true.
Where EDR Replacements Actually Break
You've picked the new endpoint platform. The harder part is the order of events. When does the old contract end? When is the new agent in prevent mode on every server? When does the MDR provider stop watching one console and start watching the other? And what do you tell your insurer meanwhile? Get the order wrong and you can have a week where a slice of laptops runs neither agent, or a renewal application that says "EDR on 100% of endpoints" when that wasn't true on the day you signed it.
This EDR migration checklist rests on two rules: no endpoint loses protection, and no statement to an underwriter goes out of date. The ten steps:
- Reconcile a complete endpoint inventory.
- Negotiate overlap licensing on both contracts.
- Brief your broker and insurer before you sign.
- Pilot both agents side by side.
- Tune the new policy in detect mode.
- Turn on prevention, one policy group at a time.
- Roll out in rings, with exit criteria for each ring.
- Remove the old agent in batches, keeping tamper protection on until the last moment.
- Hand over MDR monitoring with a parallel run.
- Build a dated evidence pack for renewal.
A mid-size company with 200 to 2,000 endpoints should plan on roughly 60 to 90 days from start to finish. Large estates with VDI, OT-adjacent systems or isolated segments often need closer to 120.
Count Every Endpoint and Pay for the Overlap
Step 1: inventory. Your old EDR console is not your inventory. It lists only devices that already have the agent, so it can't show you the ones that never got it. Pull lists from your directory (Active Directory or Entra ID), your MDM (Intune, Jamf, Workspace ONE), your CMDB, every cloud account, and a network scan. Then reconcile them. Devices that appear as "managed but not protected" are your real starting gap: forgotten build servers, kiosk PCs, contractor laptops, autoscaling cloud workloads. Fix those first, or the migration will carry them over.
Step 2: overlap licensing. Don't let the old contract expire before the new platform is fully deployed. Ask the outgoing vendor for a 60 to 90 day extension, billed month to month at your current rate. Ask the incoming vendor for a ramp schedule so you pay for seats only as you deploy them. Many vendors will agree to both, especially if you raise it before signing rather than after.
| Cost driver | Typical 2026 range (USD) | What to negotiate |
|---|---|---|
| Old EDR extension | $3–$9 per endpoint/month | Month-to-month at your existing rate, not list price |
| New EDR licensing | $4–$12 per endpoint/month | Ramp pricing tied to the deployment schedule |
| MDR service | $8–$25 per endpoint/month | A parallel-monitoring start date; onboarding fees waived |
| Migration services | $5,000–$40,000 one-time | Fixed scope with named deliverables |
These are typical 2026 market ranges. They vary by contract size, region and reseller, so confirm them with vendors. Here's a hypothetical: a 500-endpoint company paying $4–$8 per endpoint per month for a 90-day extension spends roughly $6,000–$12,000 on overlap. That is often comparable to or less than a cyber policy retention, and it's the cheapest protection against a coverage gap during the switch.
Brief Your Broker Before the New Contract Is Signed
Step 3. Cyber underwriters ask specific questions about endpoint controls. Which EDR do you use? What share of workstations and servers does it cover? Is it monitored 24/7? Is tamper protection on? Some carriers also favor certain EDR and MDR vendors or price differently depending on which you use. A vendor swap can change your terms, sometimes for the better.
Disclosure rules differ by market. In the UK, the Insurance Act 2015 places a duty of fair presentation on commercial buyers. Many European civil-law markets have their own pre-contract disclosure regimes. In the US, misrepresentation rules are set state by state. In many Asian and Middle Eastern markets, the declaration on the proposal form carries most of the weight. Your policy wording may also require you to report material changes during the policy period. Ask a licensed broker how your wording and jurisdiction handle this. Don't guess.
Two practical rules apply almost everywhere:
- Don't schedule cutover close to renewal. A window of 30 days either side is a sensible buffer. If you can't avoid it, describe the migration and its dates accurately in the application rather than answering for the finished state.
- Choose the telemetry data region before you deploy. EDR telemetry includes personal data. GDPR in the EU and UK, India's DPDP Act, Saudi Arabia's PDPL and Brazil's LGPD can all affect where it should sit, and many vendors can't easily move a tenant to another region later.
Pilot Both Agents Side by Side Before Anything Scales
Step 4. Pick a pilot ring of about 5% of devices that reflects the whole estate. Include IT staff, a few finance users, developers, at least one server per operating system, a VDI golden image, and some remote laptops on poor connections. Install the new agent in detect-only mode alongside the existing one.
Problems usually come from two kernel-level agents both trying to block the same activity. Following each vendor's published guidance, add the other agent's processes and directories to each product's exclusions. If you're moving to or from Microsoft Defender for Endpoint, check how Defender Antivirus switches to passive mode. Servers often need explicit configuration rather than switching automatically.
Test the deployment method too, not just the agent: installs through Intune, SCCM, Jamf or Ansible, installs through authenticated proxies, offline installers, and re-imaging. As a reasonable exit gate, require about two weeks without crashes or boot failures, sensor health at or near 99%, and sign-off on every business-critical application.
Tune in Detect Mode, Then Turn On Prevention by Policy Group
Step 5. Run the new platform in detect or audit mode for two to three weeks across the pilot and the first rollout ring. Review every high-severity alert. Some are real findings the old tool missed. Most will be noise from backup agents, RMM tools, developer build chains and line-of-business software.
Build your exclusions from scratch. Copying the old exclusion list is the most common mistake in an endpoint protection migration, because it carries years of shortcuts into the new platform. Watch for these red flags:
- Whole-directory exclusions such as temp folders, user download paths or entire developer drives
- Exclusions by file name only, with no path or signer
- Exclusions nobody can explain
Step 6. Split policies into groups: standard workstations, privileged admin machines, developers, and servers by role. Switch each group to prevent mode separately. Turn on tamper protection, behavioral ransomware blocking and, where licensed, device control and rollback. Confirm each group with a controlled, benign test using the vendor's test tools or a standard test file. Check two things: that the action was blocked, and that the alert reached your MDR provider's queue. A block that pages nobody is only half a control.
Roll Out in Rings With Hard Exit Gates
Step 7. Move the rest of the estate in rings. Measure coverage daily as devices with the new agent in prevent mode divided by total reconciled inventory, not by the number of devices in the new console.
| Ring | Scope | Typical duration | Exit gate |
|---|---|---|---|
| 0 – Pilot | ~5%, mixed roles and OS | 2–3 weeks | No crashes; apps signed off |
| 1 – Early | IT and low-risk workstations, ~20% | 1–2 weeks | Prevent mode; help-desk tickets at baseline |
| 2 – General | Remaining workstations, ~50% | 2–3 weeks | Nearly all of ring healthy |
| 3 – Servers | Production servers and cloud workloads | 2–4 weeks, in change windows | Each server role tested by its owner |
| 4 – Stragglers | Remote, legacy OS, special-purpose devices | Ongoing until closed | Every device covered or in the exceptions register |
Check legacy operating systems early. Older Windows Server and Linux versions are often supported with fewer features, or not at all. Anything the new agent can't protect goes into a written exceptions register with compensating controls, such as network isolation or application allowlisting, plus an owner and a retirement date. Underwriters generally accept documented exceptions far more readily than undisclosed ones.
Uninstall the Old Agent Without Disabling It Early
Step 8. Tamper protection is built to stop exactly what you now need to do: remove the agent. Most vendors, including CrowdStrike, SentinelOne, Sophos and Trend Micro, use maintenance tokens, passphrases or similar credentials for uninstalling. Get them through the old console or API, in bulk where the vendor allows it, well before the contract ends.
The biggest red flag in any EDR replacement plan is losing access to the old console on expiry day. That leaves orphaned agents you can't cleanly remove, and some may stop updating while they still sit in the kernel.
Don't switch tamper protection off across the whole estate. Script removal per batch:
- Confirm the new agent is healthy and in prevent mode on the device.
- Retrieve that device's uninstall token.
- Uninstall, then confirm services, drivers and scheduled tasks are gone, using the vendor's cleanup tool if needed.
- Record the result against the inventory.
Before you lose access, export historical detections and incident records. If a breach later turns out to have started before cutover, forensic investigators and your insurer's claims team may ask for that history.
Hand Over MDR Monitoring With a Parallel Run
Step 9. An MDR provider transition takes one of three forms: the same provider on a new EDR, a new provider on the same EDR, or both changing together. The third is the riskiest, so avoid changing both in the same quarter if you can. If you must, keep the outgoing provider watching the old agent until the last device is uninstalled, and have the incoming provider watch the new console from the first pilot device, not from cutover day.
Plan for two to four weeks of overlapping monitoring. Hand over a written package covering:
- Open incidents and active investigations
- Asset criticality, plus executives and other high-risk users
- Escalation contacts and call trees across time zones
- Pre-approved response actions, such as isolating a workstation without asking, versus servers that always need human approval
- Known-benign behaviors found during tuning
- Your IR retainer and the insurer's breach hotline or panel requirements
Then test it. Raise a controlled alert outside business hours and time how long it takes to reach a named person. Confirm too that your insurer's panel incident response firms can work with the new platform's telemetry. Most support the major vendors, but verify rather than assume.
Close Out With a Dated Evidence Pack for Renewal
Step 10. When the rollout is done, assemble the documents an underwriter or auditor will ask for and date each one:
- A coverage report: devices in prevent mode against reconciled inventory, split into workstations and servers
- A policy export showing tamper protection and prevention settings are on
- The MDR contract, its SLA, and the result of your out-of-hours escalation test
- The exceptions register with compensating controls
- Confirmation that the old agent has been removed, plus the archived telemetry export
Use this pack to update your renewal answers, and send it to your broker before the application is due, not with it. If the new platform improved your position, for example through 24/7 MDR or broader server coverage, the evidence supports a conversation about terms.
One rule decides whether you're finished. If you can't produce a dated report showing the new agent in prevent mode on every in-scope endpoint, keep the old license running. This week, run the inventory reconciliation from Step 1 and send your broker a short note covering the planned vendor change, the dates and the overlap period.
Frequently asked questions
How long should the overlap between old and new EDR licenses last?
For most companies with 200 to 2,000 endpoints, 60 to 90 days is a reasonable planning range. That covers piloting, tuning, ringed rollout and batch removal. Large or complex estates with VDI, legacy servers or several regions often need around 120 days. Negotiate month-to-month terms so you can end the overlap early if the rollout finishes ahead of schedule.
Can we safely run two EDR agents on the same device?
Usually yes, for a short period, as long as only one is actively blocking and each product excludes the other according to the vendor's published guidance. Crashes and slowdowns tend to happen when both agents are in prevent mode at once. Keep the new agent in detect mode until it has been tuned, and test servers and VDI images first.
Do we have to tell our cyber insurer we're switching EDR vendors?
It depends on your policy wording and jurisdiction. Some policies require notice of material changes during the policy period, and renewal answers must be accurate on the date you sign. Disclosure duties differ between, for example, the UK, EU member states, US states and markets in Asia and the Middle East. Ask a licensed broker to review your wording before cutover.
What if the outgoing vendor won't extend our contract month to month?
Ask for a short fixed extension, such as one quarter, or buy extra time through a reseller, which often has more pricing flexibility. If neither works, bring the start of the new deployment forward so prevent mode is live everywhere before expiry. Retrieve all uninstall tokens and export historical telemetry well before the end date.
Written by the Sentinel Underwrite Ops editorial team. Prices, features and rules change often - confirm details with the vendor or a licensed professional before you decide.