Batch, streaming and edge audiences in AEP: how evaluation works

Why this matters

Understanding how audience evaluation works is critical because the method chosen directly affects when and how users are personalised. If you pick batch evaluation for a complex rule, users won’t see updates in real time—personalisation will lag, especially during high-traffic events. This delays response to new data and reduces campaign effectiveness.

Using streaming or edge evaluation for simple rules ensures immediate updates, improving relevance and user experience. However, if you apply a complex rule to streaming or edge without checking compatibility, the rule fails to evaluate at all. This leads to incorrect audience membership and broken personalisation. Choosing the wrong method means either missing real-time opportunities or failing to support the rule entirely. The right choice balances speed and rule complexity to deliver accurate, timely personalisation.

The key ideas

An audience is defined by a rule that determines which profiles meet specific criteria. The way that rule is evaluated depends on the method used: batch, streaming, or edge. Batch evaluation runs once every 24 hours across all profiles and is best for complex rules that cannot be handled in real time. Streaming evaluation updates audience membership in real time as new data arrives, but only supports certain rule types. Edge evaluation runs on the Edge Network and enables personalization on the same page or next page, using a limited set of rule types.

The choice of evaluation method involves a trade-off between speed and rule complexity. Streaming and edge are faster and provide real-time updates, but only work with simpler rules. Complex rules that cannot be processed in streaming or edge are evaluated in batch, even if that means slower updates. So, when building a new audience, you must consider whether the rule needs immediate updates or can wait until the next batch cycle.

In short, batch is reliable for complex rules, streaming offers real-time updates for simple rules, and edge enables fast personalization on specific pages. Each method serves a different use case, and the right choice depends on your audience’s needs and rule complexity.

How to apply it

Start by defining your audience rule clearly. Ask whether the rule depends on real-time data or if it’s based on historical profiles. If it requires immediate updates when new data arrives—like a user’s latest interaction—choose streaming evaluation. This method updates membership in real time for supported rule types, enabling faster personalization.

If your use case involves personalization on the same page or next page, and the rule fits within the Edge Network’s capabilities, go with edge evaluation. It delivers faster, on-the-fly personalization directly in the browser, but only for specific rule types.

If the rule includes complex logic—such as multiple conditions, nested rules, or data sources not supported by streaming or edge—batch evaluation is the only option. It runs once every 24 hours across all profiles, evaluating all data in a scheduled batch. This method ensures accuracy but introduces a delay.

Always consider the trade-off: speed versus complexity. Simpler rules can use streaming or edge for real-time results. More complex rules must rely on batch evaluation, even if it means slower updates. Choose the method that best matches your audience’s data flow and personalization needs.

Mistakes to avoid

• Avoid designing complex rules that rely on real-time data or multi-step logic without checking whether they support streaming or edge evaluation. Such rules will fall back to batch evaluation, which introduces delays and reduces personalisation responsiveness. Always verify rule complexity against supported rule types before deployment.

• Don’t assume streaming or edge evaluation is always faster or more efficient. If a rule requires data from multiple sources or nested conditions, it may not be feasible in streaming or edge environments. Choose the right method based on both rule complexity and performance needs. Over-relying on edge for broad audiences can lead to inconsistent results due to limited rule types. Always evaluate the trade-off between speed and rule feasibility.

Quick checklist

• Identify the audience rule type: if it involves real-time data changes, streaming may be suitable; if it requires complex logic, batch is likely needed.
• Determine if personalisation is happening on the same page or next page: same-page use cases may require edge evaluation.
• Check if the rule includes complex logic or multiple conditions that exceed streaming or edge capabilities.
• Consider the required update speed: streaming and edge offer faster updates, but only for supported rule types.
• If the rule is too complex for streaming or edge, batch evaluation is the only viable option.
• Confirm whether the audience will be used for dynamic content or user targeting across sessions.
• Evaluate whether real-time membership updates are essential—batch offers consistency but slower response.
• Align the evaluation method with the data sources and profile update frequency.