Hands down that is the ugliest test I have ever seen in my life. I always recommend to take Anti-Virus tests with a grain of salt in general, regardless of the source, but this review just takes the biscuit.
Unless the author is telling me that they have reverse engineered McAfee products to determine exactly how the web filter has been implemented and works, and then also reverse-engineered web browsers to determine how their web filters work, the test is completely unrealistic and inaccurate when it comes down to the URL test. How do you know that the browsers aren't doing a database comparison before a connection is actually started without checking? How do you know the browser doesn't do its own internal checks before the security product is even aware of the connection, or at-least to the extent where it can use the received data for it to perform a scan?
Google/Firefox intercepting and blocking a malicious URL before McAfee doesn't necessarily mean that McAfee missed it, it may have identified it if Google/Firefox hadn't. If not through blacklisting, maybe through web-based heuristic analysis... I can see Intel having implemented something of the sort for their McAfee products. Plus if the data is blocked by a third-party, it prevents the AV from receiving the remaining which could have lead to a detection. Until you reverse engineer, you can't say for certain. Therefore this entire test was stupid and the "missed" URLs because of third-parties identifying first is nothing but pure, misleading speculation.
The "New" samples are VirusSign? LOL! You've got to be kidding me?? That is just stupid, but what is more stupid is how McAfee don't go through new VirusSign packs and add them to the database with a proper detection name. They should have had 100% for that, surely they have access to VirusSign paid samples as well as the free list. The more the merrier
What a stupid test.