I've touched on this in previous posts but I really believe that it is one of the most important topics that we can discuss. Of all the skills the best-run companies can have, it is pattern recognition that sets the highest functioning organizations apart. The Wharton Business School even offers an entire course dedicated to this in their Advanced Management Program. As Al Filreis, Kelly Professor in the Wharton AMP asks "Can you quickly assimilate a disparate set of data and see a pattern emerging and then anticipate the completion of the pattern?"
How does a management team know if a data point, or even a data set is an anomaly or a complete market shift?
Take for example the data set in figure 1.1.
Taken by itself it doesn't really tell us much. But if we look at the context, or "the bigger picture", we can see that it is just part of a larger story.
Seeing the context of the data, it's easy to visualize the whole pattern. In fact, this exact kind of pattern recognition is used in Wharton's School of Business:
In Wharton's AMP, the process of pattern recognition is explored by experts in history, anthropology, poetry, and other disciplines. An anthropologist, for example, can look at a single human footprint in Kenya and draw out a whole story of a people who lived tens of thousands of years ago. By looking far beyond the borders of business, the program encourages executives to take a fresh look at how they recognize patterns in their own work.
While these skills are important for executives at any point in their careers, they are particular vital to the senior managers in the AMP. "They can be good at certain content and basic knowledge but still fail because they don't know how to interpret what is in front of them," Filreis said.
In fact, Business Intelligence, and in particular Data Visualization has become interchangeable with the identification of patterns. Management teams are constantly being put in the position to have to make decisions faster, and with the fewest possible data points while still being held to high standards as opportunities present themselves. This can be challenging for several reasons:
- Finding order in conflicting and chaotic data
- Knowing when there is enough information to make a decision.
- Maintaining the ability to be neutral about the data and examine it without a preconceived notion.
- Don't be too risk averse. You can learn as much or more from failures as successes.
The simplest and often best tool most companies can use to find patterns that can affect their business is a dashboard of pertinent information about their business. Looking at many measurements in the context of how they relate to each other gives a much broader picture of what is actually happening and it means that any single data set will need fewer data points in order to accurately make an overall decision. I have generally heard that it takes a minimum of 6 data points (although in a quick search I did not come up with a reference to back that up) to make a trend. But it can take a lot more than that if not all those data points follow the specific trend. So, for example, if you are looking at a revenue graph alone, it may take a year or more to see a consistent trend when you look at seasonalities and normal fluctuations month to month. However, if you were looking at revenues, booked business, and A/R days together, you might start to pay attention if revenues and booked business fell simultaneously while A/R days grew. This could mean a business slowdown that could require immediate action.
I think that pattern recognition is a skill that every management team must master in order to optimize their business opportunities. A failure to see these patterns may mean missing the opportunity to make a course correction or, just as importantly, a lost market opportunity.