Improve Skill Performance with New Skill Metrics on Latency and Responsiveness

The conversation between your skill and your customers should happen at a natural cadence and should feel conversational. If you ask a person a question and they don't respond, or they take a long time to respond, you might want to end your conversation with that person. Likewise when your skill experiences latency or fails to provide a response from your endpoint, end-customers experience friction that can lead to a loss of engagement.

Starting today, you can monitor your skill’s endpoint latency and endpoint response so that you can stay on top of how your skill is performing right from the Analytics tab of the Alexa Developer Console. Endpoint latency depicts the amount of time your skill takes to respond to customer requests. Endpoint response tells you how often your skill was unable to respond to customer requests due to endpoint unavailability. 

With the new analytics dashboard metrics, you are now able to drill down into latency issues that may be deteriorating your customer experience. For example, if you notice your latency P90 is taking longer than expected, you may have performance issues to resolve within your code or database calls, you might consider adding caching with API Gateway, or you may need to upgrade your AWS tier or your compute platform plan where you host your HTTPS service endpoint tier to increase performance. After you determine that your performance is optimized, you might consider using the Progressive Response API to provide customers with an auditory loading indicator. This will not reduce latency, but may aide in improving the customer experience while waiting for a response.

For more advice on reducing latency, please visit our blog on improving Alexa skill performance. 

Starting today, you can access the performance metrics in the analytics dashboard. You can navigate to it by clicking “Analytics” from the top navigation bar on the Alexa Developer Console. From there, you can select “Endpoint Latency” and “Endpoint Response” from the menu on the left. The new latency and response metrics are filterable by region, time interval, and skill stage. You can also change by which period the data is aggregated. This can help you pin point a skill operational issue.

Improve your skill performance by studying and taking action against the new performance metrics in the analytics dashboard today. Visit our docs for more information and to get started.