A reciprocity failure compensation recommendation - whether from a data sheet or an "app" - is always somewhat subjective.
When reciprocity fails, it tends to affect the shadows much more than the highlights. When you choose how much to compensate for that failure, your choice may be highly dependent on what part of the image (shadows, mid-tones or highlights) that you are concentrating on.
The differences between the data sheet and the apps may very well be due to the differences in what they are measured from.
Try both recommendations in a number of different lighting scenarios. You may find yourself using both of them, just in different circumstances.