The top three commandments on Decision Analytics for Decision Analysts, from a group of close to 90 decision professionals gathered for the DAAG Conference in Austin, are: Embed Decision Analysis, Focus on Results and Incorporate Human Judgment. Embedding decision analysis involves the application of the principles of decision analysis to everyday work processes of an organization without much fanfare around the term “decision analysis” or the consulting process behind it. Decision analysts tend to care more about decision quality and not so much on results. However, decision makers care more about results, and therefore, decision analysts need to de-emphasize the distinction between decisions and outcomes. Decision makers can exclusively rely on analytics and get anchored to incremental improvements. By incorporating human judgment, they can break out of this trap and leverage the unique human ability of synthesis.
In an action-packed session involving speakers from companies as diverse as Chevron and Google, we learned that:
- “Analytics” has eclipsed all other keywords searched for on the INFORMS website.
- Big data won’t solve our innovation challenges and we need human judgment.
- Incorporating human judgment with Data Analytics leads to powerful marketing insights.
- Decision Analysis tools produce powerful analytics that drive a day-to-day supply chain operation.
- Embedded decision analysis involves integrating the principles of decision analysis in regular work processes and shifting our focus from delivering decisions to delivering results.
- Applying decision analysis in other domains without wearing the “decision analysis” badge works well.
Bill Klimack, Decision Analyst at Chevron, reported on the enthusiasm around analytics in the INFORMS Analytics Conference, which had wrapped up in San Antonio just before DAAG began in Austin. He shared with us two charts – the first was of trending words on the INFORMS website, where Decision Analysis was at the bottom. Replacing “Decision Analysis” with “Analytics” completely eclipsed everything else (on a normalized scale). In this backdrop, Bill established a compelling need for Decision Analysts to acknowledge the massive interest in analytics to drive better decision-making.
Akshay Rajwade, Product Manager at Google, cautioned against losing sight of the importance of human intuition. He pointed out that big data can get us trapped into an incremental innovation mindset as opposed to focusing on breakthrough innovation. For example, analyzing tons of data around email usage would never have produced the insight around the need for social media. Akshay made the distinction between analysis and synthesis, and noted that synthesis was a particularly strong human ability that was at the heart of innovation. He pointed out that a computer can play a sequence of notes, but it takes a human mind to produce beautiful music with it.
John Busbice, Founder of MIDA, illustrated how marketing, like big wave surfing, is now about leveraging big data to go to where the big waves are. Combining data and judgment provides better results than just using one or the other for marketing mix decisions. Statistical methods can help us find patterns but they are limited to what we can see in the data. Human judgment can be creative and logical but it is limited by instinctual biases. Using the strength of one to counteract the weakness of the other makes for a powerful combination. Combining Decision Analysis with data analysis allows us to systematically incorporate human judgment. By incorporating human judgement as prior probability assessments, John does Bayesian updates and develops marketing mix models that learn from data.
Vikram Naidu, Finance Leader for Print and Digital at Chegg, shared how he uses the tools of Decision Analysis, in particular, the Tornado diagram, to identify key metrics that drive the value of a book purchased on their site. These metrics then drive supply chain operations, with targets that are monitored and fed back into Vikram’s model. Vikram has created a decision process where, without much fanfare around Decision Analysis, he is able to focus the organization’s conversations around uncertainties that really matter. Thus, he focuses on basic conversations around clarity on the problem to be solved and uses modeling to help key stakeholders appreciate uncertainty in their operations.
David Matheson, CEO of SmartOrg, provocatively asked why Decision Analysis is not as prevalent as accounting. The Decision Analysis community is preventing the rapid spread of Decision Analysis principles with their fixation on the traditional consulting approach. If not traditional consulting, then what? David defined “Embedded Decision Analysis” as the use of the principles of Decision Analysis within everyday business processes of an organization.
This involves using systems that rest on the foundations of Decision Analysis to combine data from people and data from other systems and produce key metrics around prioritization of work that can fit right into the project phase gate. David then shared an example of a company that performs 3000 decision analyses for each of its projects every year with just such a system, utilizing ranges instead of single numbers for market forecasts, using Tornado diagrams to focus attention on the most important factors, etc. Embedded Decision Analysis is then a producer of decision analytics that moves the organization forward. Embracing this view of decision analytics shifts our thinking dramatically, as shown in the attached table. By merely shifting our own thinking, decision analysts can create lots of new opportunities for themselves.
The thread of Embedded Decision Analysis ran through Vikram, David and Clint’s talks. The session ended with attendees capturing the “Commandments of Decision Analytics” and voting on it. After compiling all the inputs, three themes emerged, as shown in the graphic. The three dominant themes were “Embed Decision Analysis,” “Focus on Results” and “Incorporate Human Judgment.”
Specific comments for the theme “Embed Decision Analysis” included:
- start focusing on improving the business process of organizations to incorporate key principles of decision analysis
- stop trying to sell “Decision Analysis consulting”
- speak the language of the client and not in “strange tongues”
- be humble
- be open to learning, instead of just teaching.
Specific comments for the theme “Focus on Results” included:
- Focus on the deliverable
- Don’t present DA as only process, but focus on results too
- Connect with client results
- Show how this will deliver results
- Focus on application
- Wrap DA with results
Specific comments for the theme “Incorporate Human Judgment” included:
- bringing Decision Analysis into Data Analytics
- focusing on framing before we get into analytics
- not taking the human out of the loop
- having conversations on uncertainty about prior knowledge
- not take big data as “the solution”
- not let data distract us from the opportunity
Taken together, these three commandments say that Decision Analysts have a vital role in the world of analytics, but we need to take off the “Decision Analyst” badge.
My gratitude to my co-chair Jack Kloeber in helping put this session together. The videos here have been produced by Onder Guven, a PhD student at Stanford in Decision Analysis. Onder is studying how organizations adopt new business processes (in particular, DA), and needs as much help as he can get. You can help open doors for Onder by talking to him about how your organization is adopting Decision Analysis. We invite you to write to him (onderguven at gmail.com) and offer help.