Endpoint protection is the area where most organizations believe they are covered because something is installed on the machines. Yet classic signature-based antivirus and EDR differ by an order of magnitude in what they detect and what they can do about it.
The difference
| Criterion | Classic antivirus | EDR |
|---|---|---|
| What it looks for | Known malware signatures | Suspicious behavior |
| Against zero-days | No protection | Partial protection |
| What it can do | Quarantines files | Stops processes, isolates machines |
| Investigation | None | Retrospective search |
| Fileless attacks | Cannot see them | Sees them in the process chain |
A large share of today’s attacks bring no file with them. They use legitimate system tools instead (command shells, scripting engines, administrative utilities). Signature-based protection cannot see these, because there is nothing to recognize.
What to look for when choosing
Whether it covers your entire device fleet. Windows, macOS, Linux servers, and legacy systems if you have them. A solution that covers only workstations is half-finished.
Whether it offers retrospective search. If a new indicator of compromise comes to light three weeks from now, can it tell you whether it ever appeared in your environment? That requires retaining telemetry, not just alerts.
How many false positives it produces. Only a pilot can measure this. Ask for a 30-day trial in your own environment, not a demo.
Whether it can feed your SIEM. EDR is one of the densest signal sources you have. If its data cannot be exported, it stays isolated. We covered this in our article on SIEM deployment.
Who responds to the alerts. This is the most important question, and it is not a technical one. If there is nobody to watch them, you need a managed service (MDR) rather than a standalone product.
Deployment steps
1. Inventory. How many endpoints do you have, and which operating systems do they run? Deployment coverage of 85 percent means the attacker works on the remaining 15 percent.
2. Monitor mode. For the first 2-4 weeks the EDR only logs and alerts; it does not block. This is when you find out which legitimate tools trigger alerts.
3. Tuning. Every false positive needs a decision: an exception, a rule refinement, or acceptance. Keep exceptions narrow, documented, and periodically reviewed.
4. Gradual blocking. Start with high-confidence detections, then widen the scope.
5. Response process. Who receives the alert, what they do with it, when they escalate, and what happens at night and on weekends.
What often gets forgotten
Checks after deployment. Coverage has to be reviewed regularly in the EDR console. A new machine, a restored system, a forgotten server: something always slips through.
Tamper protection. The first thing an attacker does is shut down the EDR. Turning on tamper protection and alerting on any attempt to disable it are baseline requirements.
Servers. Many deployments stop at workstations, yet the attack continues on the servers. EDR running on the domain controller and the file server is disproportionately valuable.
Legacy systems. Where no agent can be installed, you need compensating controls: strict segmentation and closer network monitoring. In industrial environments this is the usual situation.
What to measure
- Coverage: on what percentage of the inventory it runs and reports
- Mean response time from the moment an alert appears to the action taken
- False positive rate after tuning
- Tamper protection: whether it is active on every machine
If coverage is not above 98 percent, every other metric is theoretical. Our IT security services include designing and deploying an endpoint protection architecture.