Personalisation used to mean adding a name and a friendly tone. Now it includes recalling history, predicting intent, and adjusting workflows by customer type. That can drive better outcomes, but it can also introduce risk when it is inconsistent or ungoverned.

The signal is uneven experiences. High-value customers get great service in one channel and generic service in another. AI agents personalise based on incomplete context. Agents hesitate because they are unsure what they are allowed to reference.

Operational personalisation starts with rules. Define what data can be used, where it comes from, and how it should be presented. When the system controls personalisation, it feels seamless and stays compliant.