by Jesse Wolfersberger | June 25, 2018

Perhaps the strangest job in baseball is the clubhouse attendant. The clubhouse attendant is sort of a personal assistant to the players on game days. On any given day, they are likely to go on food runs, pick up dry cleaning, pick up an aunt from the airport, make recommendations for local restaurants, clean the dirt out of a pair of cleats, or wash a car.

The clubhouse attendants are typically college kids in their early 20's, but there are some lifers who have held the role for decades. The hours are brutal -- 12 to 18 hours on game days, which include every weekend and every summer holiday. The base pay isn't very lucrative, but the tips are legendary. Each trip to the local smoothie shop is worth a few hundred bucks, and the end-of-season-thank-you tip could be upwards of five digits.

Over the course of a season, a clubhouse attendant gets pretty good at knowing the players and their preferences. Any given attendant becomes exponentially better at his job as the season goes along, to the point where the players don't even need to ask for what they want anymore. Their dry cleaning is already hanging up in their locker, starched exactly the way they like it. 

Whether the attendant recognizes it as such, he is making a predictive mental model about each player's preferences. At the scale of 25 players, a human can predict and remember preferences accurately. If that attendant had to do this for hundreds or thousands of players, it wouldn't be possible. This example is the struggle of the modern rewards program -- you can't cater to individual preferences and still scale to your whole program. Reward catalogues have never been bigger and points have never been more liquid, but programs feel less personal than ever. Artificial intelligence is here to help.

One of the key strengths of AI is its ability to make hyper-targeted, accurate predictions at the individual level, no matter how many individuals there are. In fact, the more people, the better the predictions will be for everyone, because there are more data points for the AI to learn from. 

At Maritz, we recently completed a pilot program using AI to predict reward preferences, in conjunction with our client HSBC. Our AI was tasked with learning the preferences of cardholders, then email each person and suggest a reward option to them that fit their preferences. Of the customers who redeemed, 70 percent did so in the category that the AI recommended for them. This was a huge indicator for us that the future of rewards is in AI and predictive personalization. 

Many members of incentive and loyalty programs accumulate points in their account and never redeem. There are too many choices, too many hurdles, too much friction for them to spend the time and effort to shop with their points. Instead, they either default to cash-like options or just let the points expire. In this test, our AI acted as a virtual clubhouse attendant, one who knows the customer well enough to surface the perfect reward option, all without even needing to ask. 

This pilot is not the end-game, not even close. I think it's the tip of the iceberg of where AI is going to take this industry. Rewards will probably always involve items and currency to earn, but the definition of what we consider a reward will change. In the near future, your reward program will not just be a set of options, it will be a service that helps improve your life. Through AI, we can begin building programs that help customers save time, help salespeople find high-potential prospects, and help employees improve their skill sets with targeted learning. Being in a loyalty or employee program will be like having a virtual clubhouse attendant who is there to make your life easier, works 24/7/365, and doesn't require any tipping.

Jesse Wolfersberger is Chief Data Officer for Maritz Motivation Solutions, and specializes in merging the fields of behavioral science and artificial intelligence. Contact him to discuss if you are using data in your programs to make them smarter at [email protected]