The first step is, don't read them at all. I rarely do. I just read the parts that are important, usually the introduction and the conclusion. Pay special attention to the pros and cons listed at the end of the review. Every camera made is a compromise in some way. If they weren’t they would all cost $10,000. So look carefully at the cons and decide if you can live with that particular camera’s shortcomings-- maybe it's too tall, or too short, or doesn't laugh at your jokes. Oh, sorry, I forgot, we're talking about photography.
Pay absolutely no attention to the final rating given, whether it's a number (out of 10, as they use on cnet) or some other rating system. These are all pretty meaningless. Take this review by dpreview.com of the new Ricoh R8. They give the camera a "Recommended" rating, despite saying things like this:
If you are looking for a camera to take flash pictures… stop reading now and start looking for alternatives. The R8's flash performance is pretty poor to say the least. The flash is quite underpowered to start with but even when subjects are actually positioned within its reach the flash exposure gets it wrong more often than not.
Wow. that's a pretty big flaw for a $400 camera. Dpreview also notes that the R8 has a terrible movie mode, a face detection system that doesn't detect faces, an image stabilization system that doesn't stabilize images, and performs poorly at pretty much all ISO settings, though they're only "slightly disappointing" by this last point.
Clearly, this is a camera to recommend.
This brings me to my final piece of advice: look carefully at the photo samples. Ricoh is generally a pretty good company, but most of the photos for the R8 look terrible to me, either noisy, or worse, full or noise reduction smudging out the fine details. This is particularly evident when you look at the largest sizes, which you should always do. Buying a camera is like a forensic science. The more you can see, the better.