Farmers AVP: 3 Keys to Translating Analytics into Business Impact

On the face of it, the decision to use analytics to drive business results is a no-brainer. A majority of business leaders claim to understand analytics, and they believe that their organizations currently use analytics or they plan to use analytics in the near future. In reality, though, adoption of data-driven decisioning is quite limited.

There are three things that need to come together in order for analytics to truly influence business results: 1) Asking the right question(s); 2) Analyzing the right data in the right manner; and 3) Communicating effectively.

The Answer Is Only as Good as the Question

A good question can come only from understanding where you are and where you want to go. As mathematician and "Alice in Wonderland" author Lewis Carroll said, "If you don't know where you're going, any road will get you there." The quality of analytics depends on how clearly business leaders define the problem to the analytics team. A successful analytics team needs the right-brain ability to ask intuitive questions and thoughtfully challenge conventional practices as much as it needs the left-brain logic and reasoning prowess.

Let's say we want to understand if better customer service leads to better results. This usually is viewed as a question on the relationship between customer satisfaction scores and the service quality. Satisfaction scores almost always rise with better service (see Chart 1). However, the real intent of the question should be to understand if there is business value in providing better service. "Do better-served customers stay longer with us?" The answer to this question might give you a very different perspective on the value/ROI of better service levels (see Chart 2).

Garbage In, Garbage Out

This not only is true of the data that is used for analysis, but also for the methodology used. As per Mark Twain, "It ain't what you don't know that gets you into trouble. It's what you know for sure that just ain't so."

In an effort to provide "an answer," analytics teams often resort to using the best available data, even if it is not an ideal fit. While this approach can be acceptable under certain circumstances, it usually is better to create the required data through testing. In fact, the best approach for any analytics team is to continually seed tests that will provide data for analyses scheduled for a year down the line. The "continuous test and control" mind-set also provides the additional benefit of monitoring your current strategy and presenting immediate triggers for taking corrective action.

Once we have the right data, we need to ensure that the analysis design is appropriate for the problem. A common analytical error is that of mistaking correlation with causation. We usually are looking for the latter, but most analyses end up proving and presenting the former. Decisions based on this misconception can prove quite costly.

Let's take the previous example of customer service and add a new scenario: What if customer service reps have been providing better service to high-value clients (not a bad thing), but the analysis team does not know about it? High-value clients tend to have a greater intrinsic loyalty, and therefore would exhibit higher retention. If the selective treatment information is incorporated in the analysis, or if the analysis includes a multivariate regression, it will become clear that the relationship between service levels and retention was more correlation than causation, and therefore improving service levels will have an even lower impact.

Keep It Simple

If your internal customers don't understand your message, they might seem to agree with your recommendation, but they are unlikely to act on it. Analytics groups tend to focus on sharing their research methodology, but that approach is more relevant for peer reviews or statistical publications. In the board room, the focus should be on the insights and the business results that the approach can drive. Care should be taken to use business language and to exhibit sufficient knowledge of execution challenges. This last step ensures that the audience does not take your work as an academic exercise.

Once you have appealed to the logical sensibilities of your internal customers, you also need to generate a positive emotional response. Logic may lead to conclusion, but emotion leads to action. In order to evoke emotion, we need to develop association. Instead of delivering a finished product in the end, the analytics team should partner with the business every step of the way. The sense of ownership is a powerful motivator for action.

In the end, an analytics team needs to function as an influential change agent to truly drive business results. This requires introspection, as well. Are your presentations too dense or complex? Are your insights actionable? Have you built credibility with quick wins before going on to recommend big changes? Have you thought through the operational challenges?

It is not an easy road. Remember that the more transformative the insight, the greater the time required for its adoption.

About the Author: Neeraj Arora is assistant VP of customer experience at Farmers Insurance. He has been instrumental in shaping the path of data-driven decisioning at Farmers and is helping transform the organization's thinking on customer centricity.