@RS195 Great question! The streaming audiences that are profile-only and use an Active‑on‑Edge merge policy are still evaluated by the core streaming engine, but the UI will continue to show “Last evaluation method = Batch” because that label reflects the most recent scheduled/batch run, not the continuous streaming updates. Batch evaluation is expected to coexist with streaming for these audiences; per my understanding, there is no additional hidden toggle required for real‑time qualification beyond the documented constraints on rulesets, merge policies, and ingestion.
What “Streaming” actually does
- Streaming segmentation continuously evaluates eligible audiences as profile data lands in Real‑Time Customer Profile, with qualification typically within a few minutes rather than only at the daily batch window.
- The “Evaluation method = Streaming” flag on the audience definition means the audience is eligible for that continuous pipeline, but does not suppress batch jobs; scheduled batch evaluation still runs and is what populates “Last evaluation method = Batch” in Audience Portal.
Ref: https://experienceleague.adobe.com/en/docs/experience-platform/segmentation/methods/streaming-segmen...
Why you still see “Last evaluation method = Batch”
- Audience Portal’s evaluation summary and “Last evaluation method” are tied to discrete segmentation jobs (on‑demand or scheduled) and therefore always show a batch entry as the latest completed job, even for streaming‑enabled audiences.
- Streaming updates are reflected instead via metrics like “Total qualified” plus time‑based “New audience updated” graphs, and via downstream destinations/edge lookups, not via an additional “Streaming” row in the evaluation history.
From what you described, you have aligned with the documented prerequisites:
- Active‑on‑Edge merge policy selected by the audience, which is required for streaming or edge evaluation on audiences that are to be activated on Edge.
- A profile‑based ruleset (profile‑only) using an XDM Individual Profile schema, which is explicitly called out as eligible for streaming segmentation when the ruleset pattern matches the supported streaming rulesets.
A few non‑obvious constraints can silently force a definition into batch‑only, even when “Streaming” is enabled:
- Rules that resemble “single event with no time window” or “single event with > 24h window” on experience events are now always evaluated via batch (Adobe recently tightened this and calls out that such new/edited definitions will no longer be streaming).
- If you use segment‑of‑segment patterns that combine batch‑only audiences with streaming rules, only the explicitly streaming ruleset will run continuously; the rest will refresh on batch.
Ref: https://experienceleague.adobe.com/en/docs/experience-platform/segmentation/eligibility-criteria-upd...
Given your description, you should be within the eligible patterns, but it is still worth sanity‑checking that there is no hidden event‑based condition in the definition. If, however, you do not see any near‑real‑time growth even under controlled tests (single profile ingest with a trivially simple profile attribute rule), that would point to an issue with your streaming segmentation service or sandbox configuration, and would be a good candidate for an Adobe support ticket with the segment ID, sandbox, and timestamps.
