That sounds interesting - would have to look at query pricing on the geospatial stuff - was quite a bit more expensive than AirportBoards last I looked.
Metrics on expected demand per request based on day/time is an interesting idea - I might explore that a bit. I try to reduce overhead by only requesting the data I need in those requests (usually I need lots of data for future arrival/departures and less on flights that have already arrived/departed), but I suppose I can reduce my overhead by just ignoring the incoming data on the flights I don’t need - would increase overhead on your end though (for example, sometimes I need 18 hours of future arrivals/departures at LAX, but I don’t need the 18 hours of past arrivals/departures).
I need to do some more digging, but, I suspect lots of the flights that stop getting updates in my system are a result of the [ arrival time ] == [ departure time ] convention that you guys are using - any progress on that front? It was discussed here: AirportBoards FlightXML3 Missing Flight after Arrival.