@Muzchap - that made me laugh!!
Muzchap, you need to CALM DOWN a bit, fella... and while you're there, how about we start dealing in some real evidence here rather than your own bizarre rhetoric....
Fact: GfK NOP did this study off their own backs (for PR purposes you might say). The results would not have been published by the agency if they had done the research on behalf of either O2 or Apple as it wouldn't be GfK's intellectual property. And before you ask, I work in O2's marketing department, so I'm pretty confident in this statement.
Fact: 750 is a perfectly respectable sample size, regardless of the entire population 'universe'. Your own logic on this might sound right to you, but it's cobblers. There comes a point where surveying more and more people has no discernable impact on the outcome. You'd need to understand stats to appreciate what I'm talking about here. A sample of 750 out of 64 million is infinitely more reliable than a sample of 30 out of 100.
Fact: The soup analogy is absolutely spot on. You my friend are not.
Fact: No corporate 'big cheese' as you put it would stay in his/her job long if (s)he just employed agencies to tell him/her what he wants to hear. The research agency wouldn't last long either, nor would they put their name to research that was fudged to tell the right story.
Market research / polling is a totally respectable industry - it's the (ab)users of the data that give it a bad rep.