The most common data visualisation mistake

Those who understand the value of the data in business know the important role visualisation has delivering that value. The better the visualisation the more decisions can be based on facts.

Data-centric approach

The most common way to do visualisations is a data-centric approach. Everyone who has ever visualised data has done data-centric visualisations. Take the data, select the most interesting columns and do some calculations and groupings based on those. These reports are typically containing information such as “revenue per region”, “ordered quantities per customer” or “average delivery time per factory”. Take a look at any visualisation tool case example and you will see lots of examples of this kind of data-centric visualisations.

“Instead of analysing data, we should analyse business"

Data-centric thinking affects how we think about data in general. Data is often treated like it were a hidden oil reservoir. Do some drilling and you’ll find valuable information out of the data. It’s like the information were just waiting to be revealed by a competent data analyst. 

The problem with data-centric approach is that it guides us to analyse data. Instead of that we should analyse business. There is no hidden data oil reservoir to be found which automatically would result as value. Only if we attach data to business context we get valuable information. The objective of our analysis work should be to understand business, not data per se. 

Contextual visualisation

Instead of taking data-centric approach to visualisation we should use an approach which I like to call as contextual visualisation. In contextual visualisation the approach is based on understanding how information will be consumed and used in business decision making. Providing context for the data is needed to make data content meaningful and usable.

For example, knowing the revenue per region is not the information needed by managers, but understanding which of the regions the revenue has dropped the most and as such require special management attention. Or, instead of showing average delivery time per factory there’s a need to compare the actual delivery time with customer promises and find out the deliveries which have not reach the promised delivery time. 

How to do contextual visualisation

With contextual visualisation the data is provided to users by taking the context of usage into account. Terminology used in visualisation is not based on data but that of business context. Each of visualisation elements shown should be relevant to the business decisions that can be made based on the data. Colours are not used for fun but to emphasise elements which affect the decision making process.

To make any number meaningful there needs to be context for it. Showing throughput time of 6,5 hours is not much if we have nothing to compare that with. Quite often the measured data element is available directly in data (e.g. revenue) but there’s a need to find an element of comparison in order to enable making any conclusions based on data. For example, revenue of a month is only meaningful if it can be compared with budgeted revenue of that month or with actual revenue of previous month. 

Good contextual visualisation also makes data more usable by transforming numerical values into business terms. For example, instead of showing how many days late each of the deliveries have been dispatched to customers, the deliveries can be divided into two categories: “minor delay”, “severe delay”. Concept of severe delay is then based on customer contracts which state that deliveries dispatched more than 72 hours late will trigger penalty clause.

"Data only has value if it impacts decision making."

Contextual visualisation is much harder to do than data-centric visualisation. Having good technical skills is not enough. You need to also understand laws of business represented by data and what kind of business decisions can be done based on visualisations. 

Succeeding with contextual visualisation enables transforming data into business insights and making data meaningful to managers. Data only has value if it impacts decision making.

If you want to do contextual visualisation I suggest to keep these three things in mind:

• Understand who is the business decision maker utilising the visualisation and for what kind of purposes he/she uses the visualisation
• Provide elements of comparison which enable visualisation readers to evaluate whether the values are reflecting good results or bad results
• Use only business terminology and think transforming numbers into business terms

Author is a data professional who loves to transform businesses with data. Data-enabled management is what you get with good BI. Data-enabled business is what you can achieve with analytics and AI.