I've attempted to search for anything similar on Google/here before posting this, but apologies if I've missed anything.
Using a time series graph (replicated on vertical bar and line) we're plotting Unique Count of usernames over time with the X Axis sub-bucketed by Term (country). While the data returned is correct, the way in which data is ordered (within the sub-aggs and the legend) is not what I expected.
- The order of sub-buckets only seems to be based off the values returned in the first bucket. As the first bucket in our data set is an outlier (Brazil is 3rd, Mexico is 4th) it carries this through the rest of the graph, where both are clearly well down the list if you considered the entire data set
- All buckets in the time series should be considered when calculating the order of buckets