Performance & Evaluation

Model Drift

When a model's performance degrades over time as the real world changes.

Definition

Model drift occurs when the patterns a model learned during training no longer accurately reflect the world it's operating in. This can happen gradually — language changes, business processes evolve, regulatory requirements update — or suddenly, following a major event. Monitoring for drift is an essential part of AI operations, and models used for business-critical tasks should be re-evaluated periodically against current data to ensure they remain performant.

Why this matters for your business

Models used in customer service, sales, or compliance need to be regularly reviewed and updated as your products, policies, and regulations change. A model that was accurate in January may be giving wrong answers by December.

Related Terms

MLOps

The practice of managing AI models in production — deployment, monitoring, updating.

Benchmark

A standardised test used to compare and measure AI model performance.

Fine-tuning

Further training a pre-trained model on specific data to specialise it for a task.

Heard enough terminology — ready to talk outcomes?

We translate AI concepts into measurable business results. No upfront fees — you pay only when independently verified results are delivered.

← Back to glossary

Disclaimer

This definition is provided for educational and informational purposes only. It represents a general explanation of a technical concept and does not constitute professional, technical, or investment advice. Artificial intelligence is a rapidly evolving field; terminology, techniques, and capabilities change frequently. Coaley Peak Ltd makes no warranty as to the accuracy, completeness, or currency of the information provided. Nothing on this page should be relied upon as the sole basis for commercial, technical, legal, or investment decisions without independent professional advice.

Document reference: ISO_webpage_knowledge-base_glossary_v1

Last modified: 29 March 2026

Knowledge Base·Performance & Evaluation·Model Drift