Make dataset update new table without manually edit

[tanat_inc]

I have a dataset that imports tables from a dataflow, to support dynamic customer segmentation methods. When I add a new column in dataflow, it won't appear in the dataset when it runs scheduled refresh. I have to download the dataset and then modify the power query code a little before the new column appears. 

Can I make this process automated without downloading a 3GB dataset and do this trick every time I got the order to apply the new segmentation conditions?

(FYI: Putting the calculation in the dataset itself is an ineffective way in this situation due to the large size of the dataset, and how complicated multiple steps of calculation are before we can get a single segmentation parameter.)

[lbendlin · Accepted solution]

When I add a new column in dataflow

Don't do that. Unpivot your source data.

[tanat_inc]

Brilliant!!