# Event Stream Destination Latency Metrics

{{< youtube k3D4qiRFKWg >}}<br />

For data teams, managing stakeholder expectations is critical - they want to see the right data arrive on time.

For data engineers running RudderStack, a variety of factors can impact how quickly data is delivered to destinations. Some variables, like complexity of [Transformations]({{< ref "transformations/" >}}) attached to a destination, are within your control. In many cases, though, latency can increase due to external factors like sudden traffic spikes or degraded performance of a destination API.

Previously, it was difficult for data engineers to troubleshoot data delays for specific destinations.

Now, RudderStack [Org Admins](https://www.rudderstack.com/docs/dashboard-guides/user-management/#organization-roles) can see granular latency metrics for every cloud destination, giving them visibility into delivery latency and equipping them to better manage stakeholder expectations.

These metrics include:

- **P95 latency for the selected time window**: 95% of data is delivered within or faster than this time. For example, if P95 latency for the last two hours is 9 seconds, that means that 95% of events delivered over the last two hours were delivered within 9 seconds or less.
- **Latency trend**: The latency trend shows P95 latency at intervals over the selected time window. This view is helpful for isolating the timing and impact of increased or decreased latency. 

See our [P95 Latency Event Metrics](https://www.rudderstack.com/docs/dashboard-guides/event-metrics/#p95-latency) documentation for more information on this feature.

## Screenshots

{{< image src="images/releases/p95-event-stream-latency.png" >}}