Discovery: Suppressing unnecessary errors with a caching mechanism

[timwright12]

In the app, we deal with frequent upstream errors that surface to Veterans. This will periodically show as broken features. It has gotten to the point where there are rumors (via Reddit) that the app doesn't work on the weekends (?).

Recently, @lstraghalis shared two Reddit threads where there was discussion of the claims endpoints which spurred a Slack thread about how we might be able to elegantly suppress these errors if local/cached data is available. Ideally this should cut down on negative feedback and the perception that the app itself isn't working.

Reddit Threads:

• the horror
• Every time i click on a tab in the va app it does this. Is there a fix?

These are two examples, but there are many more.

In our conversations, we discovered that caching on the backend is pretty high effort, but doing so on the frontend has a much lighter lift as the frontend team has started implementing React Query, which supports offline access as well. In this ticket we would like to come out with an effort level as well as a recommendation on what data we can/should cache locally to improve the veteran experience.

Frontend and UX should work together to identify some of the more fragile and trafficked areas where we might be able to take advantage of cached data (when there's an upstream error).

Copying @mistymg on this as well since it relates to error messaging (even though it's about not displaying them)

Definition of Done

• UX and Frontend recommendations for applicable areas of the app
• LOE for each area (design and engineering updates e.g. "Last updated on....")
• Subsequent tickets for the work created

[mistymg]

I think this discovery could be part of error content documentation epic/project. Ideally, we could document all errors and then use that documentation to look at where we could use caching to eliminate some errors.