Basing policy on surveys is quite a dangerous thing to do, as surveys, despite their best efforts, remain vulnerable to all sorts of problems. Particularly, as I note, ones that talk about "sympathy", that is very much a woolly concept. You also have the constant abuse of results that twist responses to push an agenda. One such survey, commissioned (or reported, I forget which) by the Sun, reported the shocking news that 21 in 5" Muslims had sympathy for Jihadis. Now, in the first place, note the massive disparity with various of the surveys reported in your link, so that the uncertainty attached to such results is huge -- again, this is because "sympathy" is such a woolly concept that the response from the same will almost certainly vary depending on the day you ask them, or tone of voice, etc.
Nevertheless, let's allow that it's still a potentially sizeable figure. Except, that this is not what the survey asked. It wasn't asking about Jihadis, but about "fighters in Syria". Including, therefore, any one of the 30-odd separate factions, most of which are either not affiliated with ISIS or are even actively fighting against them. It's an horrific distortion of the results -- granted, of only one particular survey, but still it makes survey-based policy or opinion always highly questionable. One shouldn't take all those claims about the common-or-garden Muslim's views with a pinch of salt, so much as a bucketful.
That aside, there are still plenty of reasons to believe that a not insignificant number of, particularly young male, Muslims are disaffected and feel drawn to ISIS. But the answer to that problem in the long-term is absolutely not to continue advocating policies that just reinforce this message, that continue to paint the current state of an affairs as an "us v. them" scenario. It risks becoming a self-fulfilling prophecy.