July 13, 2009
Prospectus Idol Entry
How Much is that Pronk Bobblehead in the Window?
To read Tim Kniker's Unfiltered post following up on one of the audience's suggested topics, surf here.
When I hear the four greatest words of spring ("Pitchers and catchers report"), hope erupts in my heart. I am powerless to withhold expressing this hope to my wife, my friends, my co-workers and my relatives. Much to their chagrin, I wear all kinds of Kansas City Royals apparel to convey this pipe dream. To keep my toggery fresh, I do an annual purchase of a few items in February when I can still convince myself that this could be "their year." As a wise woman once said, "I always say shopping is cheaper than a psychiatrist."
I don't think I'm that different than most die-hard baseball fans just because I have five jerseys, four T-shirts, 2 golf shirts, a throw blanket, nine bobble-heads, a stuffed Slugerrr, a set of drinking glasses, two squares of the old Royals Stadium turf, a pair of flannel pants, a computer mouse, an autographed George Brett jersey and bat, two baseball caps, an official base from the 2003 home opener, golf club covers, a cell-phone case, two sweatshirts, and a golf towel emblazoned with my team's logo. Well, maybe that's a little outside the norm since my team hasn't been to the playoffs in 24 years. I know what you are thinking, but hey, it's not like I've purchased my eternal resting place with the crown logo on it-the main reason being that they don't make them for Royals fans...yet.
I bring up the contents of my wardrobe to illustrate that baseball fans love their team merchandise and are just as passionate about it as any sports' fans. One way for a team to accommodate their fans' apparel needs is by placing official retail team shops strategically throughout their local market. Currently six major league teams are using such a tactic.
"Our primary goal is to drive profitable sales for the organization through the Cleveland Indians Team Shops as an extension of the brand and the box office," said Kurt Schloss, Senior Director of Merchandising for the Cleveland Indians. "By better understanding the patterns of our fans we more accurately anticipate how to serve them better through our Team Shops."
In the past few weeks, I have been working with the Indians to model their local market and determine how the Indians Team Shops help expand their brand locally. With their permission, I am using their data to illustrate the demographics of the retail team shop.
Determining a Team's Local Ticket-Buying Fan Base
To understand the merchandising potential in a given geographic area (a county, zip-code, etc.), we first need to determine the likely number of fans in that area. Luckily, we don't need to reinvent the wheel for this. In a series of articles from May 2007, Nate Silver describes a model that he uses to estimate an MLB team's market in terms of both attendance and media.
One key component of Silver's market-for-attendance model is the concept of a "claim percentage." Essentially, he is trying to estimate the probability of Joe Fan who lives 40 miles away from the ballpark buying a ticket compared to the baseline of Jane Fan who lives right next door to the stadium. Silver's formula for this claim percentage is:
Claim Percentage = ((200 - Adjusted Distance)/200) ^ 2.41
Due to a lack of available information, the parameters of 200 and 2.41 are somewhat arbitrary. The 200 miles is his estimate of the maximum distance that a "normal" fan would go to attend a game, and the 2.41 is the exponent required so that the model has a fan who lives 50 miles away be 50% as likely to go to a game as the fan next door to the stadium.
With the help of the Indians data, we no longer need to be completely in the dark about these constants -- we can shine a dim flashlight on them. I used internet single-ticket sales from the Indians for the 2009 season to determine those constants better. Also, instead of using county population data, I used U.S. Census population data at the 5-digit zip code level to achieve a finer level of granularity. Given that a significant portion of the Indians' ticket sales happen at Progressive Field's windows, I allocated these ticket sales to the zip codes within 20 miles of the ballpark, with a greater percentage allocated to closer zip codes.
In the graph below, we compare three lines based on the distance from Progressive field. These lines represent the overall population (the green line), the actual ticket sales (the blue line), and Silver's estimate for ticket sales (the red line). To put all of these on the same scale, I normalized the data by dividing the value of these numbers by what occurs within 10 miles of Progressive Field. Doing this allows us to see how distance affects ticket sales and Silver's estimation compared to the population as a whole.
To clarify, I will give a few examples. The green line (population) fifty miles form the stadium is 3.8. What this is saying is that the population within fifty miles of the stadium (2.83 million) is 3.8 times the population within ten miles (0.74 million) of the stadium.
Similarly, the ratio of tickets sold to purchasers who live within fifty miles of the stadium is 3.1 times the tickets sold to purchasers who live within ten miles of the stadium. If we assumed that there was no "distance effect" than the blue line would be on top of the green line.
The red line represents Silver's approximation using his arbitrary constants. He actually nailed it pretty well. The only thing one could say is that his model (at least in regards to the Indians) underestimates ticket sales closer to Progressive Field (in a band from 15 to 70 miles), but does a pretty good job through 200 miles. The slight overestimation outside of 120 miles is likely due to entering into another team's geography (The Pirates to the South and East, the Tigers to the west as we get close to Toledo). A 50% reduction in the error between Silver's estimation and the actual ticket sales can be gained by changing 200 miles to 180 miles and reducing the exponent to 2.32.
The Team Shop Strategy
The three goals of the retail team shop (RTS) are:
To understand if a new RTS is going to be profitable, the first step is to estimate the likely revenue generated by it. We will begin the development of that model by first understanding the demographics of the RTS.
Currently, there are six Indians Team Shops (including one located at Progressive Field) throughout the Northeastern Ohio area which spread out as far as North Canton, 53 miles south of Progressive Field. The map below shows the region with the location of each of the six shops labeled by a Chief Wahoo logo and the name of the shop to the upper left. The color density of each 5-digit zip code on the map represents ticket sales per square mile, to give the reader an idea of the ticket-buying population and a visual representation of the Indians' fan base within 60 or so miles of Progressive Field.
While all teams have a merchandise and ticket outlet located at the ballpark that is opened usually for normal business hours and home games, five other teams have also embraced the RTS concept. They are, excluding the locations at the ball park:
To model the revenue generation of a RTS, one needs to first determine its effective coverage. To do this, I analyzed the transaction data of each of the Indians Team Shop locations for 2009. Each transaction has an associated zip code which was determined by the sales clerk asking the question "Can I have your zip code?" and then typing it into the system. The graph of cumulative sales of all Team Shops based on the distance that the customer lives from the shop is shown in the graph below. Roughly 35% of the revenue comes from customers who live within 5 miles of a Team Shop, 60% from within 10 miles and 80% within 20 miles.
The really interesting insight comes when we normalize this revenue curve and compare it to the population and the ticket sales curves. The graph below shows us the extreme local nature of the RTS. For example, even though the population within 20 miles from a RTS is almost 5.4 times that within 5 miles of the RTS (and the ticket buyers are 5.1 times), only 2.3 times as much revenue is generated. Essentially, the graph tells us that distance has a much more significant effect on merchandising than on ticket sales.
The logic is pretty straightforward. For ticket sales, the alternative (watching the game on television or listening to the play-by-play on the radio) is a relatively poor substitute for the game experience. For a RTS, the alternative of online sales or buying from another closer source (a sporting goods stores, big-box discount retailers) is not as poor of a substitute. If we assume that a ticket-buying fan that lives thirty miles away from a RTS is just as willing to buy merchandise as the fan that lives five miles away, the gap between the ticket sales line (blue line) and the team shop revenue line (red line) is the merchandising sales that are going to another source, but could be captured potentially by a greater density of RTS.
Similar to the claim percentage for ticket sales, it would be beneficial to model the likelihood of a fan buying merchandise from a team shop based on his proximity to it. I found that the form of Silver's market for attendance claim percentage didn't fit particularly well, and that a better fit was based on the following formula:
Team Shop Claim Percentage = (5/Adjusted Distance) ^ 1.7
Using this formula to approximate the revenue per population metric based on distance, conceivably, we can create a model that would predict the likely total revenue generated by a new RTS. This estimate would be based on the population bands around the shop, and similarly we could calculate what revenue may be lost by existing RTS as some of the population base is cannibalized by the newer store. The Retail Team Shop Impact on Ticket Sales In 2009, the Indians Team Shops continue to be a significant channel for ticket sales. The more important question is determining if the Team Shop actually leads to increased ticket sales in the areas immediately around the Team Shop. Let's begin with the hypothesis that a RTS does not lead to increased ticket sales. If people are buying a significant number of tickets at the RTS, then we would assume that internet ticket sales in the areas immediately around the RTS should be lower than expected, since RTS purchasing is locally driven. The underlying logic is that the fan who normally would buy their ticket on the internet is now buying their ticket at the RTS because the RTS is a more convenient alternative.
Using Silver's claim percentage applied to internet ticket sales and the Census population data, we can predict the expected number of tickets purchased by each zip code through the internet channel. I examined the zip codes within 5 miles and within 20 miles of each Team Shop to see if there are reduced internet ticket sales compared to what is expected.
The table below shows a percentage which is the ratio of actual tickets sold to predicted tickets sold for 2009. A value over 100% represents more tickets sold than expected.
Team Shop 5 Miles Out 20 Miles Out Great Lakes 103% 84% Great Northern 110% 106% South Park 92% 99% Summit Mall 84% 104% The Strip 102% 96% Average 102% 98%
So whether it is 5 miles out or 20 miles out, internet ticket sales are just as expected in the zip codes immediately around the Indians Team Shops, suggesting that ticket sales at the Team Shops are not cannibalizing internet ticket sales. Therefore since no cannibalization of internet ticket sales exists, we can assume that the Team Shop increases ticket sales in the immediate area higher than would be expected if the Team Shop did not exist. Remember in our first section how Silver's model seemed to underestimate ticket sales within about 15 - 70 miles of Progressive Field. Potentially this is the effect of the Indians Team Shop.
It should be noted that we have done this analysis looking mostly at one year for one team. This has allowed us to build some models that lead us to a deeper understanding of the dynamics going on, but further validation should be done with different markets, and potentially more years. There could be effects that are unique to Cleveland's underlying economic demographics, infrastructure, and team performance. For example, Cleveland is a metropolitan area that has minimal rapid public transit compared to cities like New York, Boston, and Chicago. It is conceivable that distance has a different effect on attendance in cities with expansive rapid public transit systems and that these models would result if different numbers after calibration. On another note, the Indians have had a disappointing 2009 season on the ball field and this may shift some of the numbers closer to the stadium, i.e., there is a multiplicative effect of team record and distance. The team's poor performance may have a greater attendance or merchandise impact on the bandwagon fan 50 miles out than on the bandwagon fan just a few blocks away from the stadium.
We have investigated the effects of population and distance on both ticket sales and merchandising for RTS locations. What we have found is: