For this test, all the protection packages were required to undergo two test phases each month. First, the real-world test is performed. As part of this test, the lab exclusively uses 0-day malware, e.g. originating from e-mails or websites. The still unknown threats are partially only a few hours old, yet no more than 24 hours old. In the second phase, the objective is to identify a reference set. This contains known viruses, Trojans and other threats that are no more than 2 weeks old. Throughout the test period of 6 months, the goal was to identify many types of malware: over 700 0-day malware threats and over 30,000 malware samples from the reference sets.

System slowdown rather than protection?

Most users think that good protection from attackers always has a catch to it: sacrifices in performance under Windows. To resolve these doubts, the lab evaluates the load each product places on the system. On a standard and high-end PC without a protection solution, the lab performs defined operations: launching websites, installing and starting up programs, performing downloads and copying data. The times required then serve as reference times for the further tests. The lab repeats the operations on the test PCs with each protection package and compares the times. Only a very slight load was recorded by the lab for these products throughout all tests in 6 months: AhnLab, Avira, Bitdefender, Microworld, Symantec and VIPRE Security. For this, the packages each received the highest-possible point score of 6.

A somewhat higher load, yet not too much, was measured by the lab on these products: Kaspersky Lab, McAfee, AVG, BullGuard, Avast, Comodo, Microsoft, PC Pitstop and Trend Micro. For these products, good results of 5.8 to 5.5 points were still achieved.

Only the packages from F-Secure, G Data and K7 Computing slowed the systems down somewhat, thus scoring only 5.2 to 5.0 points. But those are still good results.

Silent partner or a pain in the neck in everyday use?

Experts often speak of "false positives" when referring to security software. The term refers to applications falsely identified by security software, thus leading to falsely initiated actions such as blocking or deleting a program. Under the category of usability, the lab examines the security packages in terms of false alarms. To do so, the packages are required to scan nearly 4 million normal files, visit thousands of websites and monitor the installation of nearly 200 clean applications. The ideal result would be not a single alarm, as none of the test cases pose any danger. The packages from Avast, AVG, Avira, Kaspersky Lab and McAfee sounded either none or only a sparse few false alarms in the test. For this performance in the test category of usability, they earned 6 points each. With just a slightly higher number of false positives, the following packages received 5.8 to 5.5 points: Microsoft, AhnLab, Microworld, Trend Micro, VIPRE Security and Bitdefender.

The average score of the lower third in this test category still reached 5.3 and 5.2 points respectively: this involved the packages from BullGuard, G Data, K7 Computing, Symantec and Comodo. F-Secure achieved only 4.7 points, and PC Pitstop 3.2 points.

Top result: 12 finishers with 17 to 18 points

In the endurance test from March to August 2018, the lab evaluated a total of 18 protection packages for Windows. During the test, 12 of the 18 products tested averaged 17 to 18 points – those are top results. Still, it deserves to be mentioned that Avira Antivirus Pro was the only product to achieve the full 18 points.

The lower half of the field ranges between 16.8 and 15.7 points. But even that is not a bad result in a test series conducted for 6 months straight. Only PC Pitstop with 13.7 points needs to improve its quality for the future.

And the test table resolves the question as to whether Windows Defender as free standard protection is sufficient: While 17.1 points is great performance on average, it could be even better.