The most important thing I know about Analytics is that no-one agrees what it means

Tony Rose over on Decision Support Analytics had this post where he asked for submissions in a particular format – the one used in the title.

I will post a couple of others this week as I have time. Here goes then with the first one:

The most important thing I know about Analytics is that no-one agrees what it means

Everyone is talking about analytics these days – analytic decision making, competing on analytics, analytic applications, adding analytics to applications, Google Analytics, web analytics yadda yadda yadda. The trouble is they all mean something slightly different – one persons analytic application is another’s dashboard.

Turning to Mirram-Webster, I find a long definition with a couple of interesting items along with links to analytics and analysis. Here’s a selection:

• analytic – of or relating to analysis or analytics; especially : separating something into component parts or constituent elements
• analytic – treated or treatable by or using the methods of algebra and calculus
• analytics -“the method of logical analysis“
• analysis – a method in philosophy of resolving complex expressions into simpler or more basic ones
• analysis – an examination of a complex, its elements, and their relations

In the world of technology and business, analytics gets used in at least 5 ways:

1. To describe the use of an data analysis tools – reports, OLAP, dashboards – or as a synonym for “business intelligence” or”decision support”
2. To quality an application – an analytic application – to imply it leverages data and analysis of that data as a core capability
3. To describe techniques for analyzing data – analytic techniques – these cover a wide range from simple to complex including data mining techniques, predictive analytic techniques and visualization techniques.
4. To describe automated tools or workbenches based on those techniques – an analytic workbench for instance – often aimed at mathematically inclined users.
5. To label an executable algorithm or set of rules that have been derived from data using an analytic technique – an executable analytic

One of the consequences of this is that one must carefully qualify the word analytics if one wishes to be precise:

• I often talk about “executable analytics” because I am talking about the representation of analytic insight in a way that can be executed by a program or service.
• “Business analytics” is perhaps a better way to discuss the overall category where one wants to include reporting, data mining and more.
• An “analytic application” is surely not just a set of pre-configured dashboards but an application that makes analytically-driven decisions.
• Competing on Analytics (the topic of Tom Davenport’s book) could involve competing by understanding one’s data better, competing by doing analysis faster or more accurately or competing by embedding analytic decision making into one’s systems. All are possible but one should be clear what one’s approach is going to be.
• When discussing analytic techniques, one needs to specify what the output of such techniques will be – models, reports, graphs or visualizations.

And so on. Analytics is too broad a term to be used lightly – make sure you know what you mean by it.