Data-Driven Decisions: Why You, Your Customers, and Your Partners Don't Make Them

SHERIDAN HITCHENS Sheridan Hitchens What is the objective of our work in data? How do we help our organizations achieve success? We help make better decisions - Who? - Executives, Staff, Customers, Partners, Regulators - How? - More accurate/certainty, Faster, Cheaper Example - View Dashboard

- Make Decision to increase marketing spend

- Improve Revenue Homo Economicus People make Utility Maximizing Decisions Eight types of Behavioral Biased Coin Heads Or Tails? 60% 40% #1 Outcome Bias We judge the quality of the decision by the outcome that occurred The Price of Hot Coffee #2 Anchoring We have a tendency to anchor our estimates on numbers that may not be relevant Guess The Price of the Flight! Route: Las Vegas -> London (LHR) Return Airline: British Airways Fare: First Class Unrestricted Booked: 30th January Leave: Monday 12th Feb Return: Monday 19th Feb

Your 95% certain confidence interval

Low End: ? High End: ? The Cost of the flight: $23,812.38 #3 Estimation Bias We are way too over-confident in our ability to make good estimates College Basketball Tickets #4 The Endowment Effect We value things that we own more than things we don’t I give you a $50, but you have to play my game!

Game 1 Game 2 A B A B Keep $30 Flip a coin for Lose $20 Flip a coin for the whole $50 the whole $50

43% chose to gamble 61% chose to gamble #5 Loss Aversion People prefer to avoid losses rather than make the equivalent gain Silver Option: Our Standard set of Services Buy $10

Gold Option: Which option do Our Premium set of Services we choose? Buy $20

Platinum Option: Our Uber-premium set of Services

Buy $50 #6 The Goldilocks Effect We are generally more likely to choose a middle option when presented with 3 or more choices Your company’s 401K Plan #7 Nudge Positive Reinforcement and Indirect Suggestions can have meaningful impacts on decisions Woo Hoo! Correlation!

Source: http://tylervigen.com #8 or Paternicity We have a tendency to find patterns in data and ascribe meaning to them What can we do? # 1 Training Systematic training to develop awareness around #2 Cultural Change Calling out our the biases in real time in our meetings #3 Automate Evaluating whether the decision should even by made by a human Sheridan Hitchens [email protected] https://www.linkedin.com/in/sheridanhitchens/