LONDON – IT industry is awash with talk of Big Data and in-memory engines. I believe we should be putting our effort into producing simple and useful data that is presented in a format that helps people communicate and make quality decisions.
Companies have invested heavily in Business Intelligence systems for many years (Gartner Research Reports BI Market Revenue Hit $12.2bn £7.83bn in 2011), to aid decision making, but do they really get value from the information they collect and present?
There are commonly three key trends in the creation of BI reports:
1 – Manual comparison
Many management packs’ have pages of tables of numbers that require readers to manually compare one month, region, or product with another to try to identify trends, patterns or outliers. This is both time-consuming and hard work.
2 – Confusing charts
Where charts are used (usually pie or doughnut) the eye is naturally poor at comparing areas and angles, so the reader usually struggles to make proper comparisons and must refer to the legend.
3 – Over-complex presentation
Some report or presentation creators use the latest flashy 3D dials and animated components. These may look impressive but often merely prove a distraction and rarely enhance communication.
The end result is reports that make life hard for those that need to make insightful decisions; consequently information is either ignored or takes too much time to interpret.
Most of the BI tools we use today tempt us into making one of these three mistakes described above.
A measured approach
To find the best way to present data, you need to understand what crucial information needs to be presented and how it can be used:
– Define what metrics will help you manage my team or role better
– Decide what is the most important metric, or the one the makes the most impact
– Establish which metrics you can improve on
Once you know what the crucial numbers are, you need to determine how they can be used:
– Decide whether it is the trend or the precise value that’s important
– Establish whether you need see the variance against a comparison or a target
– Categorised, grouped or aggregated the data to provide the best overview
Once you have a clear understanding of how the data could be used, you’re in the ideal position to actively design and present the chosen metrics in the simplest and clearest way possible.
Simply the best
– Simple line and bar charts are often the most effective way to show most types of management information
– Line charts are better for looking at trends over time
– Bar charts are applicable for category comparisons
Depending on the problem, there are some less well known techniques:
– Bullet charts are good for a compact comparison between actual figures and targets
– Spark lines are appropriate for providing an overview of a trend
– For more complex problems, tree maps, heat maps and bubble charts can be effective
These techniques for visualising data are based on the methodologies of industry gurus such as Stephen Few, a thought-leader, writer, lecturer and researcher in Data Visualisation.
With this approach some of our own customers have reported:
– Improvements in user adoption
For example, a Property company developed HR and safety dashboards which proved a great success story of global take-up.
– Clearer presentation of data
One Higher Education institution was able to show 10,000 data points in one Executive Dashboard that was clear, uncluttered and action-oriented.
– Successful Trend identification
A manager of a number of shopping centres was able to discover new trends in the performance of individual retailers over time.
(Peter Hopwood, is UK Manager at Altis Consulting, specialising in business intelligence and data management)