The 15% "typical use" failure rate is the failure rate for people who use condoms most of the time. I'm disinclined to call it a failure of condoms myself that they don't work when people don't use them, though perhaps it could be considered a defect in the method that people apparently have such trouble using them all the time. Do note that birth control research does not involve researchers observing people in their bedrooms to see how and how consistently they are using their methods. The research is based on people's reported use, so since those reports are obviously not perfectly reliable, the 2% "perfect use" failure rate almost certainly is also partly (perhaps mostly) due to occasional failures to use the condoms.
In IT, if people don't use your software correctly, it's called "Bad user interface design." In business, if people don't like your product enough to actually use it, it's considered your responsibility to make a better product next time. Most people are blaming the condom users, but I think we can take the outside perspective instead. Instead of "shoulding" the condom users, let's criticize the product:
You have to remember the product when? This is kinda bad timing to remember stuff, you know?
They have to carefully concentrate
P/S/A: There are single sentences which can create life-changing amounts of difference.