Data drift
A change in the statistical profile of the data a model receives compared with the data it was trained on, such as a new customer mix or a changed input format. Data drift can degrade model performance even when the model itself is unchanged.
Why it matters
Drift is often the earliest warning that a model may be becoming unreliable, especially when true outcomes arrive slowly. Monitoring inputs lets teams investigate before performance visibly declines.
Not every drift matters. Effective monitoring focuses on the features that most influence predictions and links alerts to someone who can decide whether to retrain, adjust or pause the model.
In practice
For example, after a UK retailer launches a new product category, its demand model starts receiving items with no sales history. Drift monitoring flags the change in input profile, prompting the team to route new items to a separate method until enough history accumulates.
Where Rodan fits
Rodan monitors input data and model behaviour in production for systems built through AI and Decision Systems.