Why organizations fail to create value from analytics, part 2

You've got data, you've sussed out facts, you've turned them into insights. Your next task: Align the behavior of all employees that are responsible for the desired outcome.

There are two key components that capture the process of creating value from analytics. I explored the first – discovering opportunities from analytics – in my previous post, and in today’s post I’m going to look at the second component: the operationalization of insights to create tangible economic value.

So what exactly does it mean to operationalize an insight?

It means to align the behavior of all employees that are responsible for the desired outcome, which involves three things:

  • Ensuring that individual employees act in a way that delivers the desired outcome from the insight
  • Measuring and monitoring the actions and outcomes of their actions
  • Providing the employees with reliable and timely feedback on their actions

Let us take for example what happens if a call center analyst discovers that customer satisfaction is the highest when customer calls last two minutes. Any calls that last less than two minutes or more than two minutes result in significantly reduced customer satisfaction. The immediate conclusion from this insight is to try to standardize all calls to two minutes. But a simple marching order to all call operators to do this will not work! Some employees may just hang up at two minutes, for example, claiming that the call was disconnected. And while they technically ‘adhere’ to the policy, in doing this they also drive overall satisfaction down.

This example shows the fundamental difference between insight and operationalization that underlies the reason why analytics often fail to deliver economic value:

  • First, an insight cannot be reliably operationalized as just a written or verbal policy. Without ongoing operational measuring and monitoring, the feedback loop is not there and the expected behavior cannot be standardized. Hence, the expected outcomes will have either unintended consequences (such as hanging up) or a high degree of variation.
  • Second, and most importantly, an insight reveals a fact – the fact that satisfaction is highest at two-minute call duration. But the insight by itself tells us nothing about the correct incentives and measurements to change the behavior of the people responsible for the outcome in the desired direction. In other words, the insight does not come bundled with a recommendation on how to change human behavior! As stated earlier, a policy to keep calls to two minutes can simply result in hanging up. And while this meets the objective, it also causes more harm than good. Hence, the measure has to include not only timing, but also satisfaction for each call and average satisfaction per employee relative to the peer group in order to prevent hanging up. None of this information was contained in the insight. Hence managers have to think how to convert an insight into an operational behavior driver.
  • Third, since operationalization aims to standardize outcomes by changing behavior, it is best done through applications and not through policy. Policies can be ambiguous and misinterpreted. Applications, on the other hand, measure precisely and provide feedback on the job while the employee is making decisions. Building the application, however, is not trivial. If not done right, operational feedback can be as ambiguous as a verbal or written policy, but in a different way. For example, publishing the insights dashboard (visualization) to call center employees is not an operationalization because the insight visualization does not contain the measures that drive the desired behavior. If a manager shares an analyst-produced scatterplot that shows that satisfaction is the highest at two minutes, that does not do any good for the employees, who will be as confused about how to interpret the data and what to do with it. They need an information app that provides on-the-job measurement and decision guidance, such as the duration of each call, the satisfaction with each call, whether their overall satisfaction is trending up or down compared to their peers, and more.

The good news is that app development has become much easier. Thus, managers can take the insights generated by analysts, think of the desired outcomes, the measurement and the drivers to change and standardize behavior, and, then work with the analysts and IT to create and publish an infoapp (information app) for on-the-job decision-making. Only then can the managers monitor the outcomes across all employees on an operational dashboard.

The creation of value from analytics links three components in the BI platform: the analytic tools to perform analysis and discover insights, the tools to convert those insights into infoapps for employees to make decisions on the jobs, and dashboards for managers to monitor whether behavior and outcomes are aligned.

This article is published as part of the IDG Contributor Network. Want to Join?

ITWorld DealPost: The best in tech deals and discounts.
Shop Tech Products at Amazon