Complete the testing – i.e. of the aggregation process as well as the extraction pipeline. Testing exists and has been done for the rest of the extraction pipeline but not for the final step of aggregation. My feeling is that there are probably some issues buried in there, even if the aggregation process runs fine and the results seem plausible. Only detailed systematic testing can reveal this but I’m afraid it will take several days of tedious work to set up the ‘expected’ data.
the original from @scanhamman: