When saving a bookmark with specific groups selected in a dynamic group by aggregated table, the bookmark "remembers" the columns that you have chosen. More specifically, it remembers the column numbers that you have chosen - this can be proven by looking at the tenant xml of a dashboard with one of these bookmarks. The problem is that these column numbers within the saved bookmark xml do not change even if the respective insight is later updated and a new column is added to that insight. This creates a problem when a field is added or removed within the table before the grouping columns that were saved in the bookmark. This is because the bookmark does not adjust for the new column which should increase the column count (or decrease if a column was removed). For example: We have a dynamic group by table that is grouped by year month and date. As a user, I have grouped this table by date only and saved a bookmark with these changes. Within the bookmark xml, you can see that this column index is [3]. Now, my expectation is to always come back to the dashboard and click the bookmark and see the insight change to be grouped by date. But if a change is made to the table as described, that will not be the case. In this example, lets say that the developer adds a 4th column between year and month called "quarter". The column index of the "date" column has now changed to [4]. However, the bookmark does NOT update its column index. So when the user comes back and clicks their bookmark, instead of it being grouped by "Date [4]", it is incorrectly grouped by "Month [3]". This needs to be "improved".
