A dry run simulates what will happen when you normalise your dataset. Because a dry run works with a sample of rows rather than your full data, it's considerably faster than normalising your dataset for real.
You can use a dry run to spot and resolve problems with your data. After doing a dry run, you'll probably find you have warning showing against some of your rows. Typically, whole groups of warnings will have a common cause - a mismatch between your data and InfoSum's Global Schema, which you will be able to resolve using mappings or transformations.
By resolving a group of warnings, doing another dry run, and repeating the process as necessary, you can improve the quality of your normalised data and therefore the value of your reports.