I disagree. Sure, use their recommendations for a starting point, but that doesn't mean you should stick to that CG. A perfect example is the Bandit, where I and many others have found that the aircraft is much better behaved (e.g. much easier to flare) with a CG rearward of the manufacturer's recommended range.

Also, the assumption that the manufacturer has done a lot of work to get the correct CG can be completely invalid depending on the manufacturer involved. I know of a number of cases where manufacturers went ahead and released a kit/ARF with an incorrect CG because the prototype was lost before extensive CG testing could be done, and they didn't want to wait until a 2nd or subsequent prototype could take up the slack.