Hold'em Manager and Poker Tracker are incredibly important if you know how to use the data. Instead of guessing or misremembering an opponent's tendencies, you can have the hard data displayed as you play, and that's invaluable for decision making. Poker is a game of mistakes and a game in which you gain when your opponent loses. Zero sum games like poker are heavily based on (hidden) information, and the more information to which you have access and understand, the better your win rate should be.
However, if you do not understand the statistics that these types of programs display or if you misapply or over-value the data, then you might get yourself into trouble trying to outplay someone simply because a number "says you should."
A common example of this type of mistake occurring has to do with sample data sample size. There are several Hold'em statistics that necessitate large sample sizes to be accurate, like cold-4-betting pre-flop, because the distribution of hands you want to analyze is only 3-8% of all Hold'em hands. One percentage point in either direction is huge for your best decision. The particular statistic simply doesn't come up and isn't measured as often as a scenario like open-raise-pre-flop, which occurs quite frequently for most players, so players tend to see a number like 88% or 12% displayed on their heads up display or within the program and believe that they have grounds to make a play with confidence.
In short, if you rely only on the display to tell you how to play, then you are missing the forest for the trees. On the other hand, if anyone used that as an excuse to say that analytics were not incredible valuable would also be very short sighted.