An example: your server is handling requests to two endpoints. Endpoint A responds in ~100ms, B in ~200ms. You start caching some responses to A higher up. Now your P99 across all servers is higher even though you improved the performance. (Potentially setting off alarms on previous thresholds)

While it doesn't solve the alerts, for viewing the data like that heatmaps are amazing. But you need the raw numbers (or predefined bands for aggregation, but that limits flexibility)