Usually when listening to DX on the bands, a listener will note the DX station as well as the station the DX station is working. Without this information it can be difficult for the DX station to look up their own transmission in their log.
For example, the other day I heard a weak station on 17m CW - ET3AA, quite rare, and difficult to work from here (Thailand). He was not particularly busy, and was rather weak. I heard him going through a few contacts, but I only really heard the prefixes of the stations he was working (UA4, DL, G), as I was trying to work him myself.
I was lucky that he heard my weak signals after about 45 minutes, but if I had completely failed to work him after an hour, and sent him an SWL QSL card instead saying "I heard you on 18MHz at 1430Z on 2015-07-07", he might have thought "yeah, I was on 18MHz at that time, that's a good one". But conversely he might have thought "you heard me? So who was I working??" and not send a QSL in response.
For a QSL sent to an amateur DX station, I would expect to see as a minimum: The DX Callsign, Date, Time, Frequency Band (or better yet, actual frequency), Mode and Station Worked by the DX station. Without that the DX station can not usually guarantee that they were actually heard at that time/date/frequency/mode, as they can't cross-check it with their own log.
One exception to this is if a DX station is calling CQ and you hear them - if you were to send an SWL QSL to them, they might not even HAVE a log entry for what frequencies/dates/times/modes they call CQ. Most logging software these days is all about QSOs, and not so much about procedural stuff like "station on/off times" or "calling CQ" or "started/stopped packet software on 2m" or whatever. Back in the days of paper logs, such things were regularly logged.