We put a lot of effort into getting the metrics, KPIs, dimensions, values into the database, data warehouse, the ETL, analysis and user requirements have all been completed but we don’t always spend enough time looking at the effective visualization of this information. Missing the step to ensure that it is as insightful, intuitive, clear and easy to use. In this post I have highlighted a couple of good ways to really confuse the end user.
Firstly a caveat and my definition of some terms – the tail doesn’t wag the dog. When talking about data visualization I am making the assumption that the data is ready – the warehouse, design, ETL processes, cube, data mart or whatever is powering the visualization is configured – rule of rubbish in rubbish out applies. I am also not including data analysis either, i.e. the process of gaining understanding or discovering meaning and insight. By data visualization I mean visual representations which enable the exploration and communication of data. As such, data visualization surfaces the results of the data preparation labours (ETL, warehouse build etc) and delivers the means to perform data analysis. So with that said below are a couple of great ways to loose the investment in the back-end and the benefits.
Ways to really confuse a user:
1. Inadequate context:
This is a really good way to confuse a user – Displaying values without showing what the value means - i.e. without giving details of whether the trend is up or down, good or bad, performing better than the previous period or nearest comparative value. Also basic things like a title, a key to colours, showing units, details of how data is being aggregated if any etc – the aim in my opinion has to be that any visualization would make sense to someone seeing the information for the first time with little to no background.
The dials on the right show two types of dial: the first shows values but no indication of any context – Giving only a numeric value reduces the value of using a gauge and could be achieved by displaying the actual values to the user which would save valuable real estate. The second type of dials show an indication of the context using colour, giving the user an indication whether the values are good or bad. These could be improved further by showing the values of the same period last year or target values and maybe a trend chart or time serie
s to assist the user – balancing the tradeoff between showing enough detail without overloading the information. There is always a difficult balance to support the information whilst providing the rich detail which supports the understanding.
The grid on the left shows how much value a trend arrow can add to show the user very quickly whether the values are going up or down. Colour can be added to this to illustrate whether the trend is desirable or not. These could then be linked to lower level information such as a time series chart to spot seasonality or other influencing factors. Notice also the title of the chart which gives the context of what is being shown.
2. Choosing the wrong visualization tool for the job:
We have a powerful repertoire of tools in the kit - GIS Maps, Charts, Data Grids, Heat Maps, Radar Maps… the list of tools is always growing and offers huge benefits when used properly.
a
It is difficult to come up with definitive rules as to when these should be used, but a rule of thumb or questions that I have been using is below:
- Impact – does this technique offer adequate impact for the information being delivered? Will the users eye be drawn to the area?
- Insight – Does the visualization technique do justice to the information being presented? Will it offer the user an intuitive mechanism to analyse the data presented? Does the visualization enable the user to perform the level of analysis required?
- Intelligence – in the context of the view or solution you are creating, does the visualization make sense? Does the positioning make sense and best use of screen real estate – i.e. the user will probably read form top to bottom, left to right? Is it consistent and meet a need? Is it as simple as possible – clear and intuitive? Is this as simple for the user to interpret as possible – the user should use all their skills to interpret the information and gaining knowledge, not trying to work out if a bubble chart is bigger than another.
3. Misuse of Colour
Use of colour is another easy way to confuse the end user. Colour shouldn’t be used haphazardly and the power of using colo
urs should be considered i.e. using hot colours such as red, yellow and green which will attract the users eye compared to cooler colours. I don’t think using retina burning, bright colours offer any benefit but they are good for attracting attention. Also using the same colours repeatedly or colours with meaning associations (i.e. red is bad, green is good), will make the user associate the values with behaviours which will save on training and bringing users up to speed.
This pie chart shows how poor colour usage can detract value. By using different hues rather than distinct colours detracts meaning i.e. which one is Footballer’s Wives and Star in their Eyes?
Another point which a colleague of mine mentions is colour blindness, 5-8% in men and 1% in woman and justifies the use of visual aids to compliment the use of colour, i.e. arrows.
4. Obscuring Information
One of the easy pitfalls to fall down is to obscure the data or its meaning. Having numerous elements on the legend for a graph, such as the pie chart above can, cause the users eyes to have to move between the key and the chart while trying to interpret the meaning that is being displayed. The chart on the left shows another example, where some data is lost because it is hidden behind other values.
Another common cause of confusion is to start the Y axis above 0 for bar charts as it can obscure the actual values and make some values look larger or smaller than they actually are. Showing a second set of Y-values, i.e. percentage can be very useful but should also be careful to not cloud the meaning.
5. Inconsistency
Using the same consistent approach for all of the visualization techniques can make a huge benefit (not least adoption and training). The temptation is often to spice things up and use a variety of tools and techniques. In my opinion the key is to use what is fit for purpose and use the right approach for the requirement but where possible use a consistent approach to colours, navigation, locations of filters, layout, positioning of buttons, details and visualization techniques.
As always I would really like to hear your comments and feedback,
John