Organizations can use the power of cultural incentives and apply it to forecasting and motivate managers to be truthful while sharing their forecasts to the board
Professor Kenneth Casey Lichtendahl Jr. is an expert in probability forecasting and combining forecast – the art of helping management cope with the uncertainty of the future. His expertise and research reaches into decision analysis, dynamic programming and Bayesian statistics and focuses on assessing, evaluating and combining probability forecasts and modeling consumption preferences of individuals.
In this interaction with People Matters, Prof. Lichtendahl discusses Decision Analysis and how organizations can leverage Decision Analysis for outperformance.
You were earlier involved with your family business but you came back to academics. What brought you back to Darden and to the field of Decision Analysis?
I got enthusiastic about the field of Decision Analysis during my tenure as a student in Darden Business School in 1996. I became curious about the subject; and realized its potential when I was able to apply it formally while working with my family business after graduating from Darden in 1998. So, I wanted to pursue it academically, and hence, I did.
What exactly is Decision Analysis? How is it done?
Decision Analysis is a scientific way to make decisions. It is a formal classification of decisions and uncertainties you face, rolled out chronologically twice. You assess all the probabilities of a course of action and assign values to all of the possible consequences of a decision. That helps you decide the best decision there is.
The most common example to explain it is whether to take an umbrella or not based on the weather forecast. Overall, there can be four consequences – you take the umbrella and it rains; you take the umbrella and it doesn’t rain; you don’t take the umbrella and it rains; and neither you take the umbrella nor does it rain. The best scenario will be when you do not have the umbrella and it doesn’t rain, and the worst of all will be when it rains and you do not even have the umbrella.
This decision-making process can actually be done scientifically and empirically by using the decision tree method, which is one of the primary tools in Decision Analysis. In simple terms, decision tree is a visualization of a decision. You assign numerical values to each consequence and based on the methodology, you can ascertain the best course of action given the circumstances.
Is there a trend of applying Decision Analysis by organizations in their everyday working? What are the potential positive outcomes of adopting this?
Decision Analysis happens often, atleast informally, in corporate board rooms. For instance, in case of a new product launch, the board members share the likely sales figures of that product and the average of those estimates becomes the group’s combined forecast.
But these predictions can be substituted by formal model-based forecasts as well, which are potentially based on science. One of the leading models of Decision Analysis is the ‘Boosted trees model’, and it is gaining a lot of traction on the crowd-sourcing platform Kaggle in the US. In India, Ola Cabs, Flipkart and Amazon India are all using this model, and it reflects in their performance – they are beyond the status of unicorns and now have multi-billion dollar valuations.
The frequency of usage of a decision tree in an organization is a great indicator of the proximity of organizations to best practices in Decision Analysis. It is important to embed Decision Analysis tools in businesses as they can be determinants of business success. Chevron, for example, is one of the world’s most advanced Decision Analysis corporate culture, and the US-based company has been found to outperform many of its peers because of its capability in Decision Analysis.
How can organizations facilitate Decision Analysis? What practices help?
The uncommon practice of incentivizing for forecasting can be used to create a culture of Decision Analysis. In this system, managers are incentivized for forecasting correctly. To give an example, consider me as the CEO of a company and I intend to know the success prediction of a proposed new product. I want to know the perspectives of my product manager, operations manager and sales manager and how they predict it will perform. If these individual managers are not incentivized to tell me what their true beliefs are, then there is a high probability they might give inflated numbers to maintain the enthusiasm and evade disappointing the CEO. But if they have an incentive for correct forecasting, then in most likelihood, they will say the truth. So the idea is to set an incentive that gets them to tell the truth and give the forecast that they really believe in.
There is a mechanism known as the proper scoring board that gives people monetary incentives to tell the truth. In a proper scoring board, the realization is compared with the forecast given by the manager, and the manager is rewarded based on the gap between the forecast and the realization. If the gap is small, the reward is higher; if the gap is large, the reward is less.
The incentives can be non-monetary and cultural as well. Today there are a few companies that are incenting for forecasting, but in a less formal way. Eli Lilly is one example. In the company, scientists issue the success probabilities of drugs passing through FDA regulations. Their probabilities are then calibrated against the actual success percentage of drugs. Organizations can follow suit and study how well calibrated people are.
How can leaders leverage data sciences to make complex decisions?
To begin with, leaders need to have a strong data science team in place, and that should be followed up with effective communication with the team to leverage the large amount of data available at their disposal. Currently, the problem is that there is a huge communication gap between business leaders and their data science teams. In order to establish a common language and understanding, it is important for business leaders to spend more time with their data scientists and the same applies to the latter. The Silicon Valley giants, Google, Facebook, Amazon and Microsoft are actually indulging in this best practice. Flipkart and Ola Cabs are the Indian examples. The technical know-how of business leaders enables this, and that bridges the gap between the technical-minded and the business-minded. Legacy companies can take cue from them and leaders can self-develop their technological competency. They can resort to MOOCs, distance learning programs offered by top schools, or even take regular classes in machine learning and big data. That’s one way they can move a step closer to their data science team.