What Real-Time Personalisation Actually Means
"Real-time" in personalisation vendor marketing describes a range of capabilities — from genuine millisecond-latency content selection at the edge to session-level personalisation that updates between page loads to batch personalisation that computes overnight and applies the results to subsequent sessions. These are not the same thing, and conflating them leads to capability expectations that infrastructure cannot meet and investment in technology that does not produce the expected results.
Genuine real-time personalisation is: content selection decisions made at request time (not pre-computed), based on the most current available audience signal (not a stale profile), delivered within the page latency budget (typically under 50 milliseconds), without degrading the user experience when the personalisation system is unavailable (graceful degradation).
The Three Personalisation Timing Architectures
Batch personalisation: Audience profiles and content selections are computed offline, typically overnight or on a regular schedule. Results are stored and applied to subsequent sessions. Latency is zero at request time. Profile currency is low — profiles may be hours or days old. Appropriate for low-velocity audience dynamics and high-latency tolerance.
Session personalisation: Audience profiles are updated at session start and content selections are computed at session or page level. Latency is low. Profile currency is moderate — profiles reflect behaviour from previous sessions. Appropriate for most enterprise personalisation use cases.
Edge personalisation: Audience profiles and decisioning logic are distributed to edge compute nodes. Content selection occurs at request time, at the geographic edge, with millisecond latency. Profile currency is high if edge profiles are kept current through streaming updates. Appropriate for high-traffic, high-velocity use cases where session-level personalisation is insufficient.
Key Takeaways
1. "Real-time" describes a spectrum of personalisation timing architectures — genuine edge personalisation, session personalisation, and batch personalisation — with very different infrastructure requirements and costs.
2. Most organisations do not need edge personalisation — session-level personalisation delivers the majority of the value at a fraction of the infrastructure cost and complexity.
3. The decision to invest in real-time personalisation infrastructure should be driven by measured latency sensitivity — demonstrating that session-level personalisation is insufficient before investing in edge architecture.