Meaningless figures
This is the kind of hopelessly shallow and ultimately pointless analysis I would expect of a high school student rather than a company that really cared about efficiency. So Sunbelt's software uses 27% CPU when performing a manual scan. This is of course more CPU time than a system that uses 100% CPU for a quarter of the time. Without complete figures the results are meaningless and you are left wondering what the company is trying to hide, assuming that nothing critical has been lost in reporting.
The whole idea of measuring performance with manual scans is a naive one at any rate. If you start a manual scan do you really care what the CPU utilisation is? More important by far is on-access performance. How much longer does it take to open a given file, or start a given application, with this software over no AV software at all? How does this compare to competitive products? I refuse to listen to the claims of a company that cannot devise a fair, accurate and meaningful method of evaluating performance. I determined the limitations in two minutes: it _must_ have occurred to someone in the company over the course of the months or years they spent developing the software.


