Statistics API returning stale queue metrics despite `realtime=true`

Building a dashboard to show real-time queue health (waiting count, agents available) for our Manila operations. We’re hitting the /api/v2/analytics/queues/summary endpoint.

The docs mention a realtime parameter to force a fresh query instead of using cached data. We’ve set realtime=true in the request body, but the response seems to lag by at least 30-45 seconds compared to the PureCloud admin UI.

Here’s the curl command we’re using for testing:

curl -X POST https://api.mypurecloud.com/api/v2/analytics/queries/queues/summary \
-H 'Authorization: Bearer <token>' \
-H 'Content-Type: application/json' \
-d '{
 "dateFrom": "2023-10-27T08:00:00.000Z",
 "dateTo": "2023-10-27T08:00:00.000Z",
 "intervalType": "minute",
 "groupings": ["queueId"],
 "metrics": ["queuedCalls", "agentsAvailable"],
 "realtime": true
}'

The response comes back with 200 OK, but the queuedCalls metric is consistently lower than what agents see on screen. Is there a specific header or query parameter needed to bypass the aggregation layer entirely? Or is realtime just a soft hint for the backend?

We’re using the Python SDK v2.1.0 for the production integration, but the curl test shows the same lag. Timezone is Asia/Manila.