Deciphering Analytics

Are your dashboards helping you plan for tomorrow?


As a user, you feel empowered as you start from aggregate levels, 'Show me territory level totals for the Eastern region, now drill down to the district level, see which territories have the highest or lowest sales. Now drill into that territory and do what is called a drill across and find out what type of customers bought from that territory.' All these drills or zoom-ins using the latest version of visually appealing code and widgets all seems rational and finally you get to a conclusion. Territory A did really well, product XYZ was bought most. Segment N were the most significant customers involved. Now you try to come to a decision for next quarter’s plan. Based on sales of XYZ last year, you are going to incentivize the sales force and set a higher quota for XYZ in that region, to get more out of the territory. Additionally, you want to market to segment N the latest features of the product.

So you run the marketing campaign for segment N and you set your quota higher for product XYZ. The quarter goes by, but here is the unpleasant surprise from this data-driven rational exercise, the result of your plan is a dip in overall sales! While product XYZ maintained its higher run rate, the rest of the territory showed a dip in sales, instead of Segment N, it was Segment P who made the higher purchases – so what happened? Instead of growing sales, your sales shrunk, your sales people did not achieve the full quota set, and a different segment than who you had marketed to got the lion's share of Product XYZ purchases. Were your conclusions from last quarter correct? Broadly speaking, are the conclusions you are coming up with, by using your interactive abilities analyzing data through flashy widgets and beautiful pretty dashboards going to help you grow your sales? Probably not!

The tools you are using when you do your attractive perusing of reports and dashboards, are all descriptive analytic tools previously known as Business Intelligence (BI) tools. These are good tools to run reports on. They are good to assess what happened in the past, and also for financial reporting. However, the aesthetically pleasing interaction with data and dashboards is not a recipe for good predictive insights. The more you use the tool and understand the data, perhaps your chances at arriving at the right conclusions go up. What they are not able to do is to give you prescriptive directions for next quarter, and in a nutshell:


In order to know 'what you should do', you have to go beyond pretty descriptive analytics charts and dashboards that are giving you views on yesterday’s data. You need a predictive model that is going to determine specifically what you should be doing, it ingests all your data, runs models on it, combines unstructured and structured data and comes up with products you should focus on next quarter and what specific campaign to run. In this world of Big Data, let the machine tell you what to do, not the savvy drills and thrills of a BI interface.

Farrukh is the co-founder of Inference Analytics. Find him on LinkedIn.

University lecture small

Read next:

How Are Higher Education Institutions Using Analytics?