Oftentimes it is beneficial to compare the perf of a run to latest data to ensure no regressions and/or to observe improvements (or lack thereof).
We have all of the historical data in blobs (soon to move to DB). We should add a step that automatically fetches this data and compares latest run.