One possibility is they only catch people at T -1, or not caught at T 0, or there could be many more caught at T -X.
In the U.S. the ATF isn't disbanding just because we haven't seen manor action by supremacists. And just because we haven't seen any announcements does not necessarily mean the ATF aren't doing their work or aren't successful.
Look at it another way: if they could field let's say 100's of attackers simultaneously, don't you think they would? Or do you think they'd keep them in reserve for the next 10 years possibly being discovered? I can't rule it out, obviously, but I think the time between insertion and activation would have to be kept small to avoid losing control of the assets once placed. I'll do some reading up on this, it is an interesting question.
If they want but can't field hundreds in the EU (or other non ME/NAf regions) does that not speak to current intel and ops being successful?
Your original assertion was "But if anything the attacks prove that the whole surveillance thing isn't effective, not that we need even more of it." To me, your responses seem to indicate the opposite [however that does not speak to the future]
Sleeper cells are a very old concept. Go about your daily business and wait for a pre-arranged signal, go to a place, pick up instructions and weapons, do the deed. It could go from 'innocent dude working in a bakery' to 'hey, that's my brother that blew up 20 people' in less than 48 hours.
And that's precisely how they should do it if they wanted to stay under the radar. This recipe could be repeated just about forever without much chance of interference, and even that wouldn't matter much as long as once every year or so a group would make it through.
The window in which one should catch a would-be terrorist is between 0 and 19 years of age, not in the last 2 days before they execute a plan. That is playing very bad odds.