When you display a web page, it commonly is not fetched via a single request. There is the master request plus many cascading sub-requests. These sub-requests fetch script files, CSS files, images, and so forth. So a 5-to-1 or higher request cascade would be quite common. Now augment that when the server begins to mis-behave and frustrated users start clicking buttons and links, thereby adding even more requests to the mix.
Before you expand the thread pool, you need to find out why the threads are not being released, because apparently there are a lot of stuck requests out there. That would seem to indicate a critical application bug, possibly due to unexpected race conditions resulting in either CPU loops (that's easy enough to detect) or deadlocks (which is trickier). Or sometimes a mix of both.
Since it looks like possibly not only is the main page request hanging up, but (if numbers mean anything), the minor stuff like page resources and image fetches as well, the first place I'd check would be to see if there are any filters in the request chain, since filters are something that almost everything gets passed through.
Customer surveys are for companies who didn't pay proper attention to begin with.