2 ms·
So the OP shouldn't use pop-ups because they'll skew numbers in such a way that he can't tell if pop-ups are effective?
by emodendroket 11y ago
So the OP shouldn't use pop-ups because they'll skew numbers in such a way that he can't tell if pop-ups are effective?
- mikeash 11y agoIt's a little more indirect, but yes. Any time you prompt the user without their asking and they put up nothing of value, the false positive rate will probably be relatively high. If only a small proportion of people sign up, then a small false positive rate (compared to the total number of visitors) can dominate. Thus I don't think the numbers given demonstrate effectiveness, and you shouldn't use something that doesn't demonstrate effectiveness. I'm a little surprised this opinion has turned out to be so controversial.
- emodendroket 11y agoIf the costs of doing it are close to nothing and there's a pretty reasonable chance it is effective there is no reason not to do it. I also find your analysis less likely than the OP's.
- mikeash 11y agoThe cost is in potentially scaring other people away because you annoy them too much. I think the potential there is fairly high.
- emodendroket 11y agoBut there are ways to get a reasonable measure of that, particularly if you at one point did not have them. Yeah, sure, it's probably not good enough to get into a peer reviewed journal, but little of the data you have when running a typical Web site is.