Discussed with Anton. This is low priority for now, since there are not really performance issues with tags, and since the tags app does not have a dedicated dev team. There are sometimes regressions, so it might be good to set up tests at some point.
When/if we do:
1. A-M to write 2 common tags scenarios
2. BE dev learns carrier.io and writes tests for those scenarios
3. Run those tests in the new performance testing environment (similar to bugfest in setup and size). Tests can be scheduled or run on demand.