Monday, December 21, 2009

Beyond the Predictive Enterprise

@ebizQ James Taylor says you (or more likely your organization) should be striving to become a predictive enterprise. His definition comes from an old SPSS presentation.
  • Derives maximum value from its data assets
  • Understands its business by gaining deep insight
  • Leverages advanced analytics to predict outcomes
  • Turns this knowledge into action to optimize decision making across all areas of its operations

But prediction forms part of an important learning loop. We make predictions in order to take better actions. There is no point making predictions about customer behaviour unless this prediction allows you to create some value for the customer.

Secondly, we take actions with uncertain outcomes, in order to improve our future ability to predict. Received wisdom may tell us that the “best” colour packaging for a given product is blue, but an inquisitive organization will continually experiment with different packaging rather than settle for a “best practice” solution.

Thirdly, intelligent behaviour allows for its own likely effects. For example, if an intelligent organization drops its prices, it has already considered how (and how quickly) competitors might respond. (In contrast, a stupid organization predicts the outcome of a marketing offensive as if they don't expect other players to retaliate.)

So instead of predictive, I prefer the term anticipative or anticipatory. Anticipation means not passively predicting the future, but taking action in advance of the future. Robert Rosen defined an anticipatory system as "containing a predictive model of itself and/or its environment, which allows it to change state at an instant in accord with the model's predictions pertaining to a latter instant". Thus anticipatory contains a degree of self-awareness and self-embodiment, which allows proactive and agile action, not merely disinterested forecasting. This is an essential aspect of what I call Organizational Intelligence.

1 comment:

Manuel Martin said...

Hi Richard,

I agree with your post. I believe we should aim higher than being 'predictive'. Most of JT's work is around decision management which actually includes some of the anticipatory or operational analytics concepts that you mention.

I have started to see many instances where predictive models directly support (affect?) the outcome of many operational decisions automatically. Furthermore, feedback loops allows them to continuously learn and fine tune themselves based on the decision outcome.

Some of this has been possible by open platforms and flexible architectures. But the main challenge I have observed comes from effectively transforming the business processes to leverage this capability.