Leaving changes to random tests of the environment as in evolution does have a drawback, it needs to be done quickly without using too much of any resource. This often means to us the effort of planning and making the changes (unless of course a different method can be used?). Research is often the quickest method of finding the necessary changes you should make to optimise a system, thus speeding up the whole effectiveness of evolution. But research brings with it a number of extra processes to deal with and a matter of risk against time and money spent.
If we include research as part of the whole process (which in an already complex environment tested by others then you must) it can speed up the process but let’s consider the costs.
You would need to create a balance of research against actual work. Ideally, the two should be combined as much as possible but normally this isn’t possible. If I were to draw out a factor from instinct I would place something like 75% work against 25% research but how would you quantify such a thing.
If you enter a totally unfamiliar environment then perhaps the total initial time should be research, but how do you know how much time to spend on this. A personal mistake of mine is to spend so much time in the research area that I find after too long that nothing has got done, so the value of what I have done is very questionable.
On the other and if you have a process that works and you continue to use it without researching for new methods or new areas then you run the risk of the environment changing and the process failing. Say for example, years ago in Internet marketing if someone were to take advantage of the then very lucrative income gained from Google AdSense and remain doing it without changing their process then without any doubt their system would have by now started to fail, if not failed completely. Change is constant so research, or some evolutionary equivalent of the change process, needs to be done constantly and consistently.
The other weakness in any area of Internet marketing is also the length of the chain of work needed to accomplish any goal. It is getting longer and longer as more things need to be included to ensure your progress is effective. The biggest example of this that I can think of is in the Social web, where the use of Web 2.0 properties and systems has taken over the old simpler process of just content and linking.