# Which Endpoint Security Slows PCs Least? (November 2026) *The two labs that measure slowdown put the lightest and the heaviest small-business engines more than ten times apart.* > Which small-business endpoint security products slow down PCs least, from the only two independent measures published: AV-Comparatives' impact score in its business performance test and AV-Test's performance mark out of 6. One row per security engine, with the products that have no performance result at all, and what a small office should do when a security agent makes its machines crawl. Endpoint Index research (Data report). Published 2026-11-01, updated 2026-11-01, data as of 2026-11-01. Source: https://endpointindex.com/research/which-endpoint-security-slows-pcs-least/ ## Executive summary **Among the small-business endpoint engines we track, ESET PROTECT Core slows PCs least, with an AV-Comparatives impact score of 3.7 in the 2026-H1 business test, and Sophos Endpoint slows them most, at 49.1.** Lower is better, and AV-Comparatives refuses its Approved status to any product scoring 40 or more; in this round that caught Sophos Endpoint. AV-Test's performance mark, out of 6, separates the field far less: Acronis Cyber Protect and Avast Ultimate Business Security hold the top mark of 6.0 in its latest business round. Only 6 of the 14 engines in our index have a current AV-Comparatives performance result, so for the rest the honest answer is that nobody has measured it. Lightness is not protection: the lightest engine here is not the best-protecting one, and the heaviest is not the worst. Read this page alongside the [lab results scorecard](/research/endpoint-security-lab-test-results/), which covers protection and false positives, before you let a slow PC decide your purchase. ## Which products slow PCs least, by lab result?

Product	AV-Comparatives impact score	AV-Test performance
ESET PROTECT Core	3.7 2026-H1	5.5 of 6 2025-12
Avast Ultimate Business Security	8.1 2026-H1	6.0 of 6 2026-08
Microsoft Defender for Business	18.6 2026-H1	5.5 of 6 2026-08
Bitdefender GravityZone Business Security Premium Also Bitdefender GravityZone Business Security	26.3 2026-H1	5.5 of 6 2025-12
CrowdStrike Falcon Go	36.2 2026-H1	Not tested
Sophos Endpoint	49.1 2026-H1 Over the 40 limit	5.5 of 6 2026-08
Acronis Cyber Protect	Not tested	6.0 of 6 2026-06

One row per security engine; a lower impact score is better. The AV-Comparatives impact score measures the slowdown a product causes in everyday office tasks on the lab's test machine; AV-Comparatives withholds its Approved status from any product scoring 40 or more. AV-Test marks performance out of 6. "Not tested" means the engine has no result in that lab's business tests. "Dated" means the round is more than 18 months old. Neither figure feeds the SMB Fit Score.

The spread is the finding. The lightest and heaviest engines in the AV-Comparatives round sit more than ten times apart on the same test machine, running the same office tasks. AV-Test's mark compresses the same differences into half-points, which is why most products score 6.0 or 5.5 there. ## What do the two labs measure? AV-Comparatives runs its business performance test on a mid-range office PC, an Intel Core i3 with 8 GB of RAM and a solid-state drive, and times everyday actions with each product installed: copying files, archiving, installing and launching applications, downloading, and browsing. It also runs the UL Procyon office productivity benchmark. The impact score combines those into one number, where a higher score means more slowdown. A product that scores 40 or more cannot be Approved in that round whatever its protection rate. AV-Test gives a performance mark out of 6 in its business Windows client rounds, based on a similar set of everyday tasks on standard and high-end PCs. It is a grade rather than a measurement, so two products a long way apart on AV-Comparatives' scale can share a 5.5. Neither figure feeds the SMB Fit Score. The Fit Score weighs protection, price, operability and coverage; it does not reward a light agent, because a light agent that misses an attack costs more than a heavy one that stops it. ## Does a low impact score mean weaker protection? No, and the round shows it. Protection rates across the same AV-Comparatives test run from 98.3% to 99.8%, a far narrower spread than the impact scores, and the lightest engine, ESET PROTECT Core, is not at the bottom of that range. The products raising the most false alarms (CrowdStrike Falcon Go, with 8) and the fewest (Microsoft Defender for Business, with 0) sit at different points of the impact table too. Weight, protection and false alarms are three separate measurements, and the [lab results scorecard](/research/endpoint-security-lab-test-results/) sets the other two out. Where weight does matter is on the machines small businesses actually run: a five-year-old laptop with a spinning disk feels a heavy agent in every file copy. On a current PC with a solid-state drive the differences in this table are measured in seconds a day. ## Which products have no performance result? Every engine marked "Not tested" above has no current business performance result from either lab. That includes some well-known EDR and managed products whose vendors do not submit to the two antivirus labs at all; the scorecard explains which take part in MITRE's evaluations instead, which measure detection and not system impact. For those products the only evidence is a trial on your own machines. ## What should a small office do about a slow PC? 1. **Measure before you switch.** Note boot time, file-copy time and application launch time with the agent paused, then running. Most slowness blamed on security software turns out to be a disk or a browser. 2. **Check the agent's scan schedule.** A full scan at 9am on a Monday slows every PC in the office; the console usually lets you move it. 3. **Exclude the folders that churn.** Build tools, databases and backup targets generate thousands of file events an hour; every vendor documents exclusions. 4. **Trial the light option on the worst machine.** If one laptop crawls, install the lowest-impact product from the table above on that machine for two weeks before changing the whole fleet. 5. **Do not trade protection for speed.** A product with a current protection result and a middling impact score beats an untested light one. ## Frequently asked questions ### Which business antivirus is the lightest on system resources? In AV-Comparatives' 2026-H1 business performance test, ESET PROTECT Core had the lowest impact score among the engines we track, at 3.7. AV-Test gave its top performance mark of 6.0 out of 6 to Acronis Cyber Protect and Avast Ultimate Business Security in its latest business round. ### Does Microsoft Defender slow down business PCs? Microsoft Defender Antivirus scored 67 of 80 on MITRE's detection view in the 2024 round, which says nothing about speed; its performance figures are in the table above, with an AV-Comparatives impact score in the middle of the field and 5.5 out of 6 from AV-Test. It is neither the lightest nor the heaviest engine tracked here. ### Why did Sophos fail AV-Comparatives' certification? AV-Comparatives withholds its Approved status from any product with an impact score of 40 or more, regardless of protection. In the 2026-H1 round Sophos Endpoint crossed that line. Protection was not the problem: the scorecard shows the protection rate for the same round. ### Does EDR slow a computer more than antivirus? Not by design. EDR records activity and sends it to a console, which adds background work, but the measured impact depends on the engine rather than on whether EDR is switched on. In the round above CrowdStrike was tested as Falcon Enterprise, which includes EDR, while Sophos (Intercept X Advanced) and Microsoft (Defender Antivirus) were tested as prevention products; the table shows how each came out. ## Methodology and caveats Impact scores are AV-Comparatives' published figures from its Business Security Test of March to June 2026 (performance test in June), read from the report page on 1 November 2026 and stored in the Endpoint Index data files; the approval threshold is quoted from the same report. AV-Test performance marks are from its business Windows client rounds, with the round shown against each product. Products sharing a security engine share a row because the labs test the engine. The labs test the vendor's business products on their own machines; your fleet's hardware, exclusions and scan schedules will change the result. Neither figure is scored in the SMB Fit Score. ## Sources - AV-Comparatives, Business Security Test March to June 2026, performance test and certification criteria, av-comparatives.org/tests/business-security-test-2026-march-june/, checked 1 November 2026. - AV-Test, business Windows client rounds (performance mark out of 6), av-test.org/en/antivirus/business-windows-client/. - UL Solutions, Procyon Office Productivity Benchmark, the benchmark AV-Comparatives uses in its performance test.