All things being equal a concordancer is only as reliable as the
corpus it uses. A corpus could be British or American English only, or only scientific texts and so on. You need to make sure the corpus is suitable for what you are searching and then, assuming the software is well built, the results will be accurate.