Our team was having a problem with very slow response times from a 3rd party web service they were contacting to get some device stats. No issues on the other end, but it had already been weeks. They ask me to take a look at it.

I take a few days, do a couple of benchmarks and tests and I isolate the section of code responsible. Turns out, the method they were calling would timeout if the device was offline. We ask the vendor, and they confirm this. They tell us to call methodX to check if it was up.

After having done that, lookup now only took seconds. They were annoyed that it wasn't documented but was just glad it was fixed.