To report or not to report? That is the analytics question

null

The Tărtăria tablets, found at a Romanian Neolithic site, are thought by some to contain the world’s oldest known written text. However shortly after the archaeologist Nicolae Vlassa unearthed them in 1961, much controversy ensued. When were they were created? What did the symbols really mean? Were the tablets forgeries? 

Archaeologists are known for their academic disputes and rivalries, but they are hardly alone. After more than 20 years in business intelligence and analytics, I find the debates raging over the Tărtăria tablets to be surprisingly familiar. Essentially the tablets are reports. While the formats for reports may have changed, we’re still facing the same challenges we have for 5000 years.

The report factory can never keep up

When it comes to getting answers, most business users have become used to asking for information in the form of reports and dashboards. And with good reason. These have been the most effective tools available for conveying numerical data. It is common practice today for data analysts to run in-house ‘report factories’ to serve the many and varied needs of their business users. 

I see myriad flaws with the report factory model, however. Reports tend to multiply like rabbits in businesses. Most companies have hundreds, even thousands of reports, only a few of which are used on a regular basis. Many are just slight variations on others. There are typically several reports referring to the same data points using different terminology. When people question the veracity of data for any reason, this prompts them to request yet another ‘fresh’ report. 

Depending on the complexity of the ask, reports can take days or weeks to generate. This means it’s invariably the most senior executives who get priority service from data analysts. Even so, by the time reports are delivered, they are likely out of date. If the report instigates a follow-up request, executives can expect another long wait.

Analytics must focus on value, not reporting

As powerful as reports can be, they are only a means of visually presenting data or complex sets of data. When focusing on the artistry of reporting it’s easy to lose sight of the fact that every analytics or data initiative should be focused on solving a business problem. The real value comes with getting the right information in the hands of the right people at the right time. The report itself is not always needed.

While all this may seem obvious, the decision “to report or not to report” is hardly ever considered. But it should be. I recommend three main criteria should always be weighed up when deciding whether to create yet another report: the time value of the information, the uniqueness of the question and the point of consumption.

Time value of information

To quote the renowned computer scientist and BI expert Bill Inmon: “At the heart of being able to quantify the value of any decision support system (DSS) is the notion that information has a different value based on the time of its first availability.”    

In other words, the value of information can diminish over time. Does it really make sense to build a report for information that will eventually become worthless?  Almost always, the answer is “no”.

Take the example of a loan company that measures its health by tracking credit scores, property locations and interest rates experiences. In this case, there is no diminishing value when monitoring those data points at fixed intervals. A daily or monthly performance report, or a dashboard, is the ideal tool in this case. 

Now take the example of a securities investment company.  Unexpectedly, the UK votes to leave the EU. This company needs to know its exposure to UK securities immediately so decision makers can chart a course of action. There’s no point in building a report or dashboard to track this as this answer has diminishing value. The real value is getting actionable information in the hands of decision makers fast. You just need an answer.  

Uniqueness of question

The next key consideration is the likelihood that the same question will be asked again. This provides a good gauge on whether there’s value in building a report to answer it.

A supermarket chain might have ten payment types, ten stores, 1000 products and 10,000 customers.  Assuming a potential intersection at every point, that’s a billion permutations.  

Now introduce a time dimension, and then more than one measure.  No organisation would try to build reports for every potential intersection, and every organisation would build at least one sales report.  

The grey area is in the middle.  Should the supermarket build a sales report for each store?  Potentially.  

How about to track sales of each product category by store? Maybe.  

What about to track a drop in cereal purchases, fourth aisle, third shelf, at a single store three days ago?  No, but it happens. 

In this type of scenario, where knowledge workers need to ask minor variations on a question of the same data set or sets, I don’t recommend fixed reports. It is impractical for IT or data experts to have to build three different reports for an area sales manager to discover, for example, which are the top performing products in store X on Monday, Wednesday and Friday.

Point of consumption

Where is a knowledge worker located when they ask for information and what are they doing? At an office desk reading emails, in a call centre assisting multiple customers or delivering spare parts in transit van?

In the desk scenario, a reporting tool is perfectly reasonable. Call centre workers constantly reacting to customers won’t have time for this and will be better served by ‘push’ notifications. A van driver will need ad hoc, accurate answers to simple questions about inventory and costs sent to their tablet.

Ways of working are evolving and so must the way we consume business intelligence. The focus today is moving towards data value, less on reporting. Who knows? It might even result in less controversy in 5000 years when the digital archaeologists unearth our corporate tablets!

Christos Mousouris, Senior Director, Global Sales Engineering & Technical Alliances for ThoughtSpot

Image Credit: ESB Professional / Shutterstock