In our particular use case, columns get added regularly and we are concerned about regular spikes in MAR as a result. After chatting with your support team, this is what I have understood:

• At the moment, the addition of a new column will cause all rows with values in the new column to be counted in MAR (in our case, a new column will almost always contain data).
• There is a manual workaround, which is to Block all schema changes and when a schema change occurs, then update the model manually and do Resync (as there is no charge for doing resyncs).

Given that there is a manual workaround, it would be great to be able to use the automatic schema change handling without having spike in MAR.