Syndicated from Smart Data Collective
Some time ago Neil Raden and I did some research on analytics. It was clear as we did this that there were two main threads of analytic use in companies – risk analytics and opportunity analytics. I blogged before on the use of analytics to manage risk one risk at a time so I thought I would write about opportunity analytics.
Risk analytics are about using historical data to make a prediction about the risk of a particular customer, a particular transaction going or being bad in some way. Risk analytics help you estimate and account for the downside risk of a decision – if I get this wrong, what’s the worst that could happen? Opportunity analytics, in contrast, are focused on estimating the upside – the opportunity.
I regularly write about the importance of focusing analytics on operational decisions and their role as a corporate asset. If we think about these kinds of operational decisions then opportunity analytics come to bear on customer-centric decisions like cross-sell and up-sell decisions, or decisions to retain a customer who has called to cancel. Opportunity analytics are used to answer questions like how profitable might this customer be in the future, how profitable might they be if they accept this offer, which offer is most likely to attract them? Opportunity analytics predict response, opportunity, potential. They predict the propensity of customers to buy products, the likely profitability of a customer if they buy a particular product, which offer is likely to be most appealing to a prospect.
Unlike risk decisions, there is often little difference between good and bad opportunity-centric decisions. If a company gets such a decision right, they might increase the profitability of a customer, or retain a customer into the future. They have little exposure if they make a bad decision. While, in theory, a bad cross-sell offer might so annoy a customer that they abandon their primary purchase, this kind of negative impact is highly unlikely. With opportunity analytics, companies are trying to maximize their upside not manage their downside. A poorly made risk-centric decision can result in fraud, bad debts, theft. A poorly made opportunity-centric decision simply wastes an opportunity to increase profitability.
This difference changes the cost justification of analytics. Risk decisions have been the more common use of data mining and predictive analytics historically because the time, hardware and skills involved could be easily justified by avoiding the potentially huge downside. Opportunity analytics are growing fast, however, as the tools get easier to use and the cost of hardware and data management continue to drop precipitously. With new tools, and more readily available experience, squeezing extra profit out of these decisions with analytics is becoming more and more worthwhile. The embedding of analytics into decision-making systems for marketing and CRM is increasingly common. Because opportunity analytics are targeting small improvements, they must change rapidly to take advantage of competitive and market circumstances. This drives an ever-increasing use of adaptive analytic models, those that use automated experimentation to constantly adapt and refine an analytic model.
Opportunity analytics may not have the pay off that risk analytics do but companies should still be thinking about using their customer data to maximize the value of every opportunity.
Comments on this entry are closed.
Hi James, I’m glad you touch on Opportunity Analytics as they tend to get overlooked.
Your comment that “opportunity analytics are targeting small improvements” reminds me of some of your other comments about how companies acquire risk one decision/customer at a time (paraphrasing) and I wonder if the difference in the financial benefit of analytics might depend more on the industry/function/volume/decision targeted rather than the type specific type of analysis, i.e. risk vs. opportunity. What do you think?
Manuel
Well the volume certainly matters -more decisions act as a multiplier for the use of analytics so higher volume decisions are always going to get a better return. I do think there is a difference, however.
When each decision has a downside risk the gap between good and bad decisions tends to be larger so the total value of the decision (gap * number of decisions) is greater, often much greater. As a result you can afford to spend more money on the analytics and you need to worry more about their accuracy.
When the gap is smaller, as it is for what Neil and I called Opportunity Analytics, the total value is lower so the automated creation and evolution of analytic models is often essential if the cost of improvement is to be less than the value created.
JT