The grain in business intelligence defines the lowest aggregation for measurements of a business process. We always recommend, unless it is not technically or financially feasible, that you capture measurements in your fact table at the lowest grain possible. You can always roll-up data to a higher grain if needed without any loss of information. This is not the case when lowering the grain. You can only lower the grain by applying allocation assumptions and estimates which are dangerous to use when making business decisions. So, store the data at the lowest grain possible.

But what about analytics? What grain should I use when performing diagnostic analysis? Match the grain to the level of decisions being made by the users of the visuals. If you are producing a dashboard for an executive, provide enough grain to inform the executive of the measurements he or she wants to monitor. Be careful not to go into more detail than requested. Executives have people that work for them that are experts in detail. Equip those people with lower grain analytics. Then, when an executive needs something explained, those that work for him or her can quickly provide an explanation using the same data set, just at a lower grain with their own set of visuals.

There is one thing to always consider when selecting the grain for making decisions – don’t allow the lack of understanding by a business user to cloud the story the data needs to share about its influencers. I have seen cases where businesses make decisions with analytics at too high of a grain – sometimes because “that’s the way we’ve always done it” – and sometime because they don’t understand that there is a more refined explanation of the influencers at a lower grain. For example, a business may perform analysis based on measurements at their various store locations throughout the country. They look at metrics like profit margin and inventory turnover per store. Taking the grain to a lower grain, like sales-rep or sales-department, might provide a more accurate story of what the data is doing and therefore enable better decision making and more resulting profit. Sometimes you might have to demonstrate the impact of this lower grain analytics using visuals so the business user can see what is happening and believe that the additional detail matters. Use everything at your disposal to help the business see how best to use their data for self-improvement.

At ConradBI we are experts in grain selection. We are also experts at assisting companies in navigating the sometimes-confusing journey of business intelligence acceptance. Contact us today to get started on your BI journey.