Analytics 101 for strategy thinkers

The term 'Analytics' increasingly finds its way into my conversations with clients these days, largely unprompted. However, like the fabled elephant and blind men, it means different things to different people.

Taking up from where I left off in ‘Getting Digital‘, let’s talk about Analytics – what it is and why it matters.

As a brief recap of the context I had set, Analytics is the domain in which copious amounts of information (Such as generated by digital networks) and cheaper more capable silicon come together towards serving a business purpose. But that is not a definition but rather a description of the context of analytics.
Before I try to define what analytics means, consider the following as a simple model for decision making in an organizational context:

What is being suggested here is that decision making is triggered by some new information being provided to a person or entity. Then the entity first assimilates or understands the new information, then combines it with existing knowledge and tries to create a mental model of what the information means, which is basically some form of pattern recognition, and thereafter moves on to generating, evaluating and choosing options of what to do based on this information (Including arguably the most common choice – Do nothing) . This decision triggers some sort of action or reaction to that information.

Assuming this is a reasonable description of how decision making works typically as we understand it, let us call this the ‘Classical’ model.

Now, as we partially spoke of in ‘Getting Digital’ , some factors at work are:

1. More information available from more points, more frequently across the whole extended value chain of the enterprise
2. This information is stored digitally and is connected to a network
3. There is cheap ‘silicon’ available for processing of this information, there are ‘cloudy’ economies of scale available in the storing and accessing of this information
4. Finally, the amount of information available, in terms of both volume and diversity, is literally unmanageable for an average human mind – There is an information overload

These factors have led to the emergence of ‘Analytics’ – which is depicted in the following ‘Analytical’ model of decision making.

Essentially, silicon is taking on roles of information assimilation, pattern recognition and response actualization to different degrees.

‘Analytics’ is machine based information assimilation and pattern recognition that contributes to the traditional decision making processes in an organization, and in some cases supplants them partially or wholly.

At the ‘assimilation’ stage, analytics typically is similar to conventional ‘MIS’ and IT – using rich data sources to create business relevant metrics that assist in decision making. When we start adding ‘pattern recognition’ into the picture, we start talking about analytics being ‘predictive’ and actually starting to suggest the actions that may need to be taken. And finally when we start creating closed loops between information and action with only limited or periodic human intervention, then this starts becoming ‘Real time analytics’ and ‘Machine intelligence’ etc. One of the things that is symbiotically driving this evolution is the Cloud – as more storage and processing capabilities are available centrally with the benefits of the cloud (economies of scale, scalability etc.), it becomes easier to actually move towards real time analytics.

And just I discussed in ‘Getting Digital’, the new realities that analytics creates have the potential to change business models because the optima shifts!

Upon adopting ‘Analytics’ in the decision making, some things change dramatically:

1. The amount and complexity of information that can be assimilated into the decision making process
2. The complexity of patterns that can be understood and used
3. The speed at which this information can be turned around into action
4. The granularity at which decisions can be taken

Let me illustrate with an example. In the ‘Classical model’ you would go to a bookstore. Maybe you’re not sure what to buy – you might consult the shop owner. Based on his knowledge of older books and new releases, his extensive experience in recommending books, your inputs on what you have enjoyed in the past and the like, he could suggest a book to you and therefore complete the transaction with you.

In the ‘Analytical’ model, the same thing can be accomplished but with a difference:

1. Instantaneously
2. Using a massive database of your and other people’s past behaviors as well as demographics
3. For thousands or millions of people at once
4. Very cheaply
5. Without needing much human expertise
6. Remotely (over a network)
7. With arguably better, but certainly more predictable results
8. Highly customized to the individual customer

Voila! Amazon recommendation engine. Game changer? Certainly.

Now when we think analytics, on one hand we really need to stretch our minds to encompass all the possible points at which information is getting generated, or could be getting generated and all the potential ways in which this information could support or supplant conventional ways of finding optimal actions for the enterprise, and in fact become transformational for the enterprise as the available optimum shifts. One the other hand, we need to marry this to consumer insight, market insight and the organization's objectives meaningfully, just as in traditional decision making.

This, I believe, is the crucial bit. Too many of us don’t understand analytics, view analytics as a hammer and set about looking for nails, or even worse try to represent it as the ideal addition to the modern toolbox without really stretching our minds and figuring out how or why the optimum for the business will shift. It’s crucial to understand that the majority of work in thinking analytics is strategy thinking, not analytics thinking. At both ends of the process and throughout, the important questions to ask are strategy questions – To what ends? What information? What actions? What patterns? The technology behind analytics makes it happen, eventually, but it is no silver bullet. We do not have any true artificial intelligence to do our thinking for us - yet.

As always, the trick lies in thinking of client context and client value in new ways.

Footnote: Try googling 'Analytics primer for CEOs'. Disappointing. This is one from HBR - very 'meh' too , imho.