Conditional formatting "per rows" with pivot dimension in Table look

[Isis_Barba]

I have a Table (Legacy) visualization in Looker with a dimension (Agrupamento Mídia / Category) on the rows and Month as a pivot, which expands my measures horizontally.

I am trying to apply a heat map / conditional formatting, but Looker applies the color scale by column (vertically) by default. Because one of my categories has very high percentages compared to the others, it skews the entire color scale, making the other rows look completely flat (all red or all green).

I need the conditional formatting to evaluate the data by row (horizontally), so that the maximum and minimum color thresholds are calculated independently for each category’s monthly history. However, I don’t see any option in the Legacy Table settings to change the formatting scope from ‘by column’ to ‘by row’.

Even when I try to create a Table Calculation to generate an auxiliary metric—which normalizes the data to show whether a value is above or below the row average—I still face a blocker. Looker doesn’t give me an option to ‘color the original measure’s row based on this auxiliary calculation’.

If I apply the conditional formatting directly to the Table Calculation, it colors the calculation column itself, but I cannot use those rules (e.g., above 1 is red, below -1 is green) to dynamically format the background of my original pivoted measures.

[zainkhan23]

I don’t think you’re missing a setting here. In the legacy table, conditional formatting on pivoted measures is basically applied per column/series, not recalculated independently per row.

The two realistic workarounds are:

• move the normalization into the value you actually display, then format that field directly
• reshape the result so each category/month comparison is no longer a pivoted measure, then use a table calc or LookML field for the row-level score

If the business user needs to see the original percentage and still color it by a separate row-normalized score, that’s the part Looker tables usually don’t handle cleanly. At that point I’d treat it as a feature request, or use a custom visualization / extension if the heatmap behavior is critical.

[ruthc]

Can you not just adjust the range settings for the rule i.e. set the END to ‘Percentile’ and adjust the figure in this setting until the outliers stop skewing the formatting? You might also need to change the CENTER setting. As a rule, I always use START = min, CENTER = median and END = 95 Percentile.

Before (using default settings)

After (adjusting the range)