Drift detector: when the ticketing platform changes without telling you

When you read data from a system that is not yours, the risk is not that it goes down: it is that it changes. A field gets renamed, a format moves, and from that day your history is comparing two different things. Nothing fails, nothing warns, and the conclusions bend slowly out of shape.

Now that change gets detected on sync, normalised, and logged with its date — so when a series behaves oddly you can check whether there was a schema change underneath instead of hunting for the explanation in the business.

It is one of those pieces that only exists if you have been reading somebody else’s data for months.