It's not that hard, you just need to use it and review that.
Problem is, that takes a good couple of hours of seriously trying out the machine for seriously doing some work. The author of this article even took a couple of days, and did some serious Googling and tweaking to smooth out any wrinkles he encountered (which is the use case for anyone who buys the device because they are expensive: if they "just work" that's great, but if there's a work-around or fix, that's still better than if it "just doesn't work").
But for many review sites, it's much faster to just run some benchmarks (or just copy the numbers from somewhere), compare that to some previous models, some current/similar models, and review that. It doesn't take a lot of effort, time or creative thinking.
For one, I think it's really hard to convey succinctly how well a product works. Many reviews will put up charts and graphs with benchmarks. But what exactly does a score of (for example) 11,204 mean? The answer, to me, is VERY LITTLE. You can compare it to another phone, but if another phone has a score of, say, 10,000, does that mean this phone is \approx 10% better? Is it 10% faster in what I am going to use it for? No, of course not --- you can't measure how good a device is with a single number, nor even if you single out just its graphics capabilities or whatever slightly more specific metric you choose.
Otherwise, articles might throw more subjective descriptions in the picture. "The experience was very smooth". Maybe reading it through quickly that sounds all well and good, but how much does that tell you? It's still a very vague term simply because many of these products have so many things going on it's hard to encapsulate them in a review in a useful way.
Anyway, in this case, the article pinpoints a very particular task and examines how good the device was at that one task. I think that is already a great sign that you can get a lot more out of this than a much more general review.