Variance waterfall start value too large compared to the change

[xingstar97]

I created a variance waterfall which shows the change of the revenue by month. The issue is the start and end value is too large compared to the changes of revenue from one month to the next month. If I set the y axis as auto, the change is too small which made the chart hard to read. If I set the min and max of the y axis, when I applied filters, the range of revenue changes, probably beyond the min and max I set before. Is there a good solution for this case?

[pravinboniface]

@xingstar97 Can you try the following on Appearance -> Y-Axis -> Range -> Custom

=RangeMin(StartValueMeasure,EndValueMeasure) - fabs(StartValueMeasure-EndValueMeasure)

This should give you a very decent chart and you can tweak the formula further according to your needs.

This is an example based on Qlik Help if you want to try it out.

Load * inline [
Quarter,Sales,Budget
Q1,9,10
Q2,14,12
Q3,12,12
Q4,18,14
];
exit Script;

Bridge Dimension

Quarter

Start Value

sum(Budget)

End Value

sum(Sales)

Y-Axis Customer Range Min formula

RangeMin(sum(Budget),sum(Sales)) - fabs(sum(Budget)-sum(Sales))

[marcus_sommer]

The waterfall chart might not be suitable for such scenario whereby all graphic visuals have challenges if the range of values is quite (too) large. Either the differences between the values are too small to read or some kind of complexity needs to be added which makes the interpretation more difficult.

The mentioned complexity means to adjust the axis and/or the values. By the axis you may enable a logarithm view but often it becomes quite hard to interpret the results and/or it takes too much time as being sensible for a visual which are in generally aimed to comprehend the results on a single/second glance. If it needs much more time a table-view might be more expedient.

Another kind of adjustment might be to mix different units within the chart. In your case maybe by showing the first and last value in k-values respectively something like: sum(Revenue) / 1000 and all the month-differences are the normal sum(Revenue) - and showing this kind of adjustment very clear in the chart and within the documentation. And of course also the complexity to get such view is much higher as by a normal chart. Nevertheless such view might be misinterpreted.

A much simpler alternatives for such view might be a combi-chart with bars for the absolute revenue on one axis and the (rate) differences on a line by the second axis.