50 Most Influential: Introduction 50 Most Influential: No. 34 Ditching ’burbs for Detroit NHL brings doughnuts, signs Dunkin’ deal 50 Most Influential: No. 16 ‘Suite’ gifts, and even a few ugly ones Group builds platform for hockey award 50 Most Influential: No. 38 Alabama scores some serious bling Sports Media: NFL steps into esports
SBJ/20080623/This Week's News
Braves leave guesswork at the gate
Published June 23, 2008
Derek Schiller, Atlanta Braves executive vice president for sales and marketing, joked recently that he’d never found a direct need in his sports career for his civil engineering degree from Vanderbilt University.
But more than a decade and a half since earning that degree, and after stops with the Atlanta Thrashers and New York Yankees before joining the Braves nearly five years ago, Schiller now is ensconced in complex mathematical formulas.
The club’s ticket sales office, led by Schiller, has developed a tool it’s calling the Attendance Projection Module. The algorithm-based system combines day of the week, month of the season, game start time and Braves opponent, and spits out an attendance projection that often is accurate within 3 percent to 5 percent for any individual game. Over long stretches of time, the accuracy improves to around 1 percent to 2 percent, club executives said.
The base data are attendance figures from 2005 to the present. The algorithm involves assigning proprietary point values to each of the four factors used to determine expected turnout, resulting in an aggregate point value for each game. A higher total point value means there are factors in favor of a big turnout, and a higher attendance is then projected.
The system marks a modern-day iteration of the time-honored but not entirely accurate practice of attendance estimation that combines the hard data of advance ticket sales and group bookings and mixes it with a fair amount of guesswork.
“What we’re trying to do is assign true logic and facts to something that previously was more experience-based,” Schiller said.
The club began using the module last year, after it was developed internally by several club employees, some with math or engineering backgrounds, including ticket department staffers Ryan McFerrin and Chad Graham. With the system’s reliability quickly shown, it has taken on greater stature this year for the club, holding more influence over staffing decisions for game operations and the timing, pricing and implementation of various ticket promotions.
Operations reports written up for each game and used internally by Braves front-office staff now include both the pregame projected attendance figure from the system as well as the actual turnout. The projection for the Braves’ June 5 game against Florida, in one example, showed an expected turnout of 27,954; posted attendance for the game was 27,238.
Before Opening Day this season, Braves officials used the module to project the team’s 2008 full-season attendance at Turner Field and found that, based on the layout of the schedule, the club would see a slight drop at the turnstiles. Indeed, the club is off about 4 percent thus far and faces an uphill climb to close the gap.
The Braves now are beginning to discuss with MLB potential wider uses for the projection module, such as inserting data from all 30 clubs into the system and perhaps using it to help draft the entire league schedule with maximum possible turnout in mind. Both camps brand those talks as very preliminary, though, and senior MLB officials have not yet conducted a full review.
“We’ve talked briefly about it, and that’s about it so far,” said Katy Feeney, MLB senior vice president for scheduling and club relations. “We’ll see what happens. I do know [attendance patterns] vary a lot from city to city.”
Neither Schiller nor Feeney was aware of a program similar to the Braves’ model being in place with another MLB club or in other leagues. Details on the cost to the Braves of implementing their program were not immediately available.The Braves’ system does not directly factor in weather and team performance for either its own club or the opponents on the schedule. Schiller acknowledged those two elements can cause short-term variances from module predictions in extreme situations. To that end, some Atlanta home games this year have shown variances from the model closer to 10 percent as the Braves, like many other MLB clubs, grappled with poor early-season weather.
Schiller said those factors typically smooth out over longer periods, though. Since the system relies in part on averages of continually updating data, sustained improvement from a once-woeful team — and, in turn, improved box office appeal for that club — would ultimately manifest itself in the attendance predictions.
“This is a guide and one that definitely helps in decision making,” Schiller said. “You don’t base 100 percent of your decision making on it, but it’s no doubt a valuable tool.”