Graphs are strong instruments in which trends can be easily shown and recognized. Many different kinds of graphs make sure that there is a graph for every purpose. With this is mind, it won’t surprise you when I tell you that graphs are also being used in Competitive Intelligence (CI) tools. But, the use of these graphs differs from for instance Business Intelligence (BI) purposes. What is the difference and what are the challenges when using graphs in CI?
We all know the typical BI graphs: for instance a pie chart for market shares, a line chart for the sales over time or a bar chart for the amount of orders per week. These graphs are based upon structured data. The total amount of sales can be calculated from order lines for example. In CI tools this kind of data display is only possible when the data you’re collecting is structured data (for instance information from Google Finance or Dow Jones index). However, in the CI process analysts gather and analyze unstructured data much more often than structured data. So the question is: How can you display unstructured data in graphs?
When you’re collecting unstructured data the only measure you can display in a graph is the amount of news items. At first, I was a little sceptic about this. What can you deduce from unstructured data, what can you conclude from the amount of news? This isn’t really a strong measure, is it?
The main reason why I wasn’t very fond of using graphs in CI is the fact they can be confusing. Take a look at the below pie chart. What is your first idea when you see this chart?
This picture does not portray market shares per company, which is probably what you thought. This pie chart shows the amount of news items per company. Because we are used to see these pie charts in relation to market shares, it takes some time to explain what this graph actually tells us. I discovered while working with graphs in CI tools that the value of the measure ‘amount of news’ entirely depends upon the way you use it.
Graphs in CI can be valuable devices. Within a CI tool, categorizing your information is the key to valuable insights. You can compare this with measures in BI (or any other type of structured information). The measure ‘revenue’ in itself doesn’t mean a lot. You need to know in what period the revenue was generated, by selling which products, in which region, et cetera. By assigning this information to the measure you basically categorize it, which is exactly what we need to do with the CI information. You can use different kinds of categories: companies, market segments, strategic areas, regions, or whatever classification is relevant for you. For instance, when you categorize you news items per company, you can portray this in a graph. We’ve done this for HP, Dell and Apple and the result looks like this:
As you can see, there are a lot of peaks in the graph. These occurrences are probably interesting to investigate. This shows us immediately the main value of graphs in CI tools: they function as warning devices. Because CI isn’t exact science, these graphs are mainly about comparisons. As you can see, in September 2010 the results are peaking for both Dell and HP. Closer investigation showed this is due to the fact they were both bidding on 3PAR, a data storage manufacturing company that they targeted to buy. In general, when results are peaking for two companies this may indicate for instance (a rumour about) an upcoming merger, acquisition or joint venture. When all companies in your graph show a peak, it can also be an indication for a market specific development. We can see an example of this in the graph in June 2010. At this moment in time, there were some issues about Foxconn, an assembly company which is contracted by Apple, Dell and HP. There were a series of suicides by employees of this Chinese company, which raised questions about the working conditions. This kind of news influences the whole industry and is likely to be important for your company.
Another example where comparisons through time can be valuable is when you launch a new marketing campaign or product. With a graph like the one above you can see whether or not your online presence is growing. It is a way to measure the online effects of marketing campaigns. An example of this is the peak in the above graph for Apple in January/February. This is the moment Apple (loudly) introduced the iPad which explains the increase of news items about Apple. Exactly that is something else that this graph clearly shows: the amount of information regarding Apple is significantly higher than Dell and HP. This is not necessarily because they are bigger or better, but because they are louder and make sure people talk about them.
Graphs like the ones we’ve seen don’t show you what the news is about. But… with the right categories, you can refine your data. For instance, you can do this by using subject categories and use these in a matrix as shown below. Here we categorize the news not only by company but also by product type. You can see that Apple is mentioned most in combination with tablets and HP and Dell in relation to laptops. This may tell you something about the market segments these companies focus on, at least in the perception of the press. Because Apple recently released the iPad, it is not surprising that Apple appears a lot in the news in combination with tablets.
Another way of refining your data is to categorize the news based on sentiment. For this, I invite you to take a look at one of our previous posts: ‘Competitive Intelligence functionality: sentiment analysis‘. This way, you are able to display the amount of positive and negative news items. In this situation graphs function again as warning devices, because you can monitor positive and, maybe even more importantly, negative coverage of your company or product.
I hope I’ve shown you that graphs are of great use for CI, if used in the right way. Graphs function mainly as warning devices or for comparisons, and don’t portray an exact measure. CI remains an inexact science and the human factor in the analysis will remain important.
Competitive Intelligence functionality: sentiment analysis by Anne van den Brink
Dell Drops out of bidding war for 3PAR from New York Post
Apple’s Steve Jobs finds Foxconn deaths ‘troubling’ from The Independent