They were a diverse group united in one way - all had an entirely different story of what "analytics" actually is. From Gary Biddle of the University of Hong Kong (and a fellow Chicago grad) talking about using techniques such as EVA to incentivise behaviour and set strategy, to Hugo Walkinshaw of Deloitte talking about "Big Data" and their Analytics Institute, everyone had a different take.
So, in order to unify these views, I present the Maroon Analytics Map. This categorises the analytics world along two dimensions:
How frequently do you make decisions? Economicsts call this "short run" and "long run". It is not a specific period of time, rather an indication of the flexibility decision makers have. Short run is about tactical execution while long run is about strategic decisions. It's also worth considering the perishability of the data involved. A high frequency trading algorithm for financial markets doesn't make sense if it takes even a second to make a decision.
This is the complexity of the infrastructure, not of the mathematical model. The model is dependent on human capital - the skills and knowledge of your workforce - and also often requires deep domain knowledge. Of course, more complex infrastructure often allows more complex mathematical models (such as Monte Carlo simulations) but to be used, but it also means more data can captured, moved around and processed in a useful period of time.
Spreadsheets dominate the "low" end of the technology spectrum, sitting just above the back-of-an-envelope. So, we most often start in the bottom-right corner when someone wants to make a strategic decision. What price should I sell at? Which product gives me the best margin? What's my efficient order quantity?
Once a piece of analytics has proven itself, the next step is usually to email a copy of it to all your co-workers who start using it on a more regular basis and this is where what could have been an asset becomes at best the residue of an idea and, in the worst-case, a business-critical liability. A skilled spreadsheet-engineer, with a bit of forward thinking, can develop these spreadsheets into valuable assets, suitable for frequent, repeat use and accessible to a wider audience. That said, they are still hard to put any controls around and many institutions will see business-critical spreadsheets as an operational risk.
Spreadsheets are still able to generate valuable insights. Many institutions only talk about analytics in terms of "big data" but this overlooks the fact that, often, "good enough" insights can be generated far more quickly and cheaply with a much smaller sample size.
Above the line, we see what is possible with more sophisticated infrastructure. Stories abound about the canny insights retailers have gleaned from understanding their consumer behaviour in more detail, such as How Target Figured Out A Teen Girl Was Pregnant Before Her Father Did and also the possible problems caused when an automated stock trading algorithm goes rogue.
And finally we get to "Big Data" and data mining. When you have automated systems capturing and storing data in an accessible way, you can develop some pretty good strategic insights. It is also possible to start in this box too, but it happens less often. Collecting or acquiring relevant data can be an expensive process. Examples include IBM's "Smarter Planet Solutions", such as their Traffic Prediction Tool which drastically cut morning commute times across Singapore.
So, from mining huge sets of data to predict traffic flow through to a spreadsheet that reveals the profitability of your products by sampling a small sub-section of available data, these are all part of the Analytics Family. And now, with the Maroon Analytics Map, you can see where they live and how to get from one to the other.