Every international student office has a version of the same artefact. A sheet, one row per person, with a date column and conditional formatting that turns something amber at 90 days and red at 30.
It works. It works until one of three things happens, and all three happen routinely.
Where the spreadsheet actually fails
A date changes upstream and the sheet does not know. A programme extension is approved, a change of level goes through, practical training is authorised. Each of those alters the relevant end date, and each is a separate process. If pushing the change into the tracking sheet is a manual step, it is a step that gets skipped on a busy week — and nothing about the sheet indicates it is now wrong.
The population turns over. People arrive, complete, transfer, withdraw. A sheet built in August describes an August population.
The arithmetic is repeated by hand. Programme end date, plus authorised training, plus grace period, per person, per change. It is not difficult. It is voluminous, and volume plus manual equals error at a predictable rate.
What changes when admission has a fixed end date
Under duration of status the tracked quantity was fuzzy — status maintenance, not a single date. With a fixed admission period there is a specific date per person, and a specific number of days remaining. That is more trackable and also less forgiving: a date can be missed in a way that an open-ended admission could not be.
Which means the tracking artefact stops being a convenience and starts being the control.
What the job actually requires
One ordered list, continuously current: who is closest to a deadline. Not alphabetical, not by programme — by days remaining, shortest first.
Arithmetic done once, centrally. Programme end, authorised training, grace period, computed the same way every time rather than per-person in a formula somebody copied down a column.
Change detection. The thing a spreadsheet fundamentally cannot do: notice that an underlying date moved and that the number on screen is now stale.
That third one is the whole problem. The first two are conveniences.
Status, plainly
AUD Watch is in development and is not available to purchase. The dashboard published on our site is a demo over simulated people who do not exist, with real-data ingestion switched off. No institution's records are in it.
The date arithmetic in that demo is real — admit-until, departure, grace period, calculated as they would be for an actual population. The calculation is built. The product around it is not.
Where it stands: AUD Watch — current status and demo