The post is written using Qlik Sense and QSDA Pro but the same principles apply to QlikView. You can use the QV Document Analyzer to get field sizes if you want to start the trimming exercise in QV.
It's quite simple - if the data-set + data-model are already well optimized the data-size couldn't be significantly reduced and it's only a question of purchasing enough resources.
That's a rather rare scenario even if experienced people designed the environment because the most development projects starts more or less iterative with a focus on development-speed and functionality. Performance is secondary as far as the run- and response-times are acceptable. If it worked it won't be further improved.
Therefore I suggest to take another more careful look to the already provided hints because the aim is not to change the information else the way how the data are stored and associated and there is often a significant potential of improvement. A very classic case would be the split of a timestamp into a date and a time field - the information would be further there but the storage/RAM consumption would be much smaller.
Beside the above you shouldn't not to disregard changes to the information too easily. For example it's quite common to analyse the data of the last n periods to an atomic level, for the periods before to maybe two years it may enough to look on a daily-level and for the elder periods it may a monthly-level. Such approach could be included within a single fact-table and will decrease the sizing significantly without a loss from an analytical point of view.
