As I mentioned in a previous post I’m in the process of automating REPAMA – our competitive marketing intelligence methodology. It’s quite a task and I’ve not been able to focus as much time on the project as I would have liked, but it’s coming along.
REPAMA visualises competitive intelligence. I created the REPAMA methodology to allow me to understand infrastructure software vendors’ marketing strategies and to allow me to compare one vendor’s strategy with another. REPAMATron is the project that is automating the manual REPAMA methodology. It’s going to take a number months to work through all of the algorithms to ensure that they are true to the manual REPAMA methodology. But I thought I’d share some of the early output and at the same time address something that is bugging me at the moment – how to represent the Market Element Distribution (MED) studies graphically.
Those familiar with the manual REPAMA methodology will know that I have historically represented the Market Element Distribution charts as spider (or radar/spoke/polar) diagrams. I used Excel to produce the charts and as you can see from the chart below this results in an interesting way to represent the relative commitment/lack of commitment to each of the specific elements of the marketing strategy.
Excel-Produced Radar Chart – Example MITICOR MED Study
The greater the commitment to a specific element of marketing strategy, the longer the spoke in the chart. For example, my MITICOR study looks at the how much emphasis a vendor places on each of the 7 categories of Value Proposition that I monitor (Market, Income, Time, Institutional, Cost, Operational, Risk). Therefore if I find evidence that a vendor leads with a value proposition around reducing expenditure (Cost is the ‘C’ in MITICOR) and has a minor commitment to increasing revenue (Income is the first ‘I’ in MITICOR), then the spider chart would feature a long spoke for Cost and a shorter spoke for Income – as shown for Vendor 2 in the example chart above.
Using Excel allowed me to ensure, through manual adjustment that each chart was readable. Even if the chart featured data that occupied the same area I would be able to adjust colour, shading or line size to make it easier to read. However, now that a program is producing the chart I no longer have the luxury of a pair of human eyes to fettle the appearance.
The result is a spider chart that can be a little busy and difficult to read as shown below.
Radar Chart – Example MITICOR MED Study (Automatically produced by REPAMATron)
In addition the charting library that I’m using doesn’t support radar charts with smoothed (spline) lines. As a result the radar features angular lines which again makes it slightly more difficult to read than it’s Excel smoothed-line cousin. (As an aside, Excel 2010 also has a problem with smoothed lines on radar charts).
So I’m looking at alternative charts that still show the relative commitment to different elements of marketing strategy but perhaps show them a little more clearly. Here are some of the ones I’ve looked at.
Spline Chart – Example MITICOR MED Study (Automatically produced by REPAMATron)
100% Stacked Column Chart – Example MITICOR MED Study (Automatically produced by REPAMATron)
100% Stacked Area Chart – Example MITICOR MED Study (Automatically produced by REPAMATron)
Line Chart – Example MITICOR MED Study (Automatically produced by REPAMATron)
Stacked Bar Chart – Example MITICOR MED Study (Automatically produced by REPAMATron)
100% Stacked Bar Chart- Example MITICOR MED Study (Automatically produced by REPAMATron)
Stacked Column Chart – Example MITICOR MED Study (Automatically produced by REPAMATron)
The style that I am currently favouring is the Spline Chart (as shown on the right). I think this really clearly shows the different strategies of the vendors in the study and the resulting spline line produces a signature shape for that vendor’s strategy.
That said in early testing I have had some feedback that this sort of chart should not be used to show discrete data values in a series, but instead is better used to show the flow between different data points – i.e. identifying the delta between previous and subsequent data points.
So if I’m not going to use the spline chart then I think I’m leaning toward the Stacked Column Chart (as shown on the left). So I will take more soundings and think on it some more.
In order to tune the natural language process engine at the heart of REPAMATron, I’m going to focus on an area of the infrastructure software market that I know well – the ESB segment. I must stress that the charts and tables that I’ll post in this and future blog entries will be Beta versions because the required algorithms and statistical analysis are not fully developed.
REPAMATron is currently designed to automatically build a series of PowerPoint slides so in subsequent blog posts I will share some PowerPoint output whilst the algorithms are tuned.