Almost every crimesolving technique and power is aimed squarely at making it easier to identify and round up criminals who are either stupid or who forgot to account for _all_ of the evidence which might be collected.
Smart criminals are always going to be harder to catch, because they're smart. But they're also a vanishingly small percentage of the criminal element. Eventually, in order to get away with a crime, luck aside, a person is going to have to be so smart that there's not really much chance of catching them at all. Even then, if the numbers of smart, successful, lucky criminals are reduced sufficiently by the percentage slivers new processes and technologies bring, that means that more resources can be brought to bear per case, and specialist crimesolving resources and teams (including things like the FBI in America) will be able to take on a greater proportion of the top cases.
As an example, if the FBI can handle, oh, ten thousand cases a year (completely guessing this number), and you're a smart successful criminal amongst 20,000 others like you, you still have a 50% chance of having the FBI after you. If new processes allow the regular police to track down, apprehend, and charge 5,000 of those smart criminals, your chance of being on the FBI's radar go up to 67%.
(Yes, I know the FBI doesn't operate like that. Replace it with any specialist resource. The point being that the more crims that can be nicked by the plod, the better the chance that the remaining ones will come to the attention of the big guns.)
Really, the smartest criminals are the ones committing crimes for which are are no laws yet. Even if they're caught, they can't be charged with what they actually did. And the best of the best make sure they have enough influence at the right levels so that they're protected from law enforcement in the first place.