On Aug. 7, 2007, Barry Bonds of the San Francisco Giants needed one more homerun to break baseball’s all-time career record. Tickets to this potentially momentous game at the Giants’ home field, AT&T Park, were surprisingly cheap.

It was a weeknight game against the Washington Nationals, a relatively weak opponent. According to the team’s three-tier pricing structure, those factors added up to a low, face-value ticket price.

"I think a ticket in the upper deck was $10," says Russ Stanley, Managing VP, Ticket Services, San Francisco Giants. "And yet, it was the hottest ticket we’ve ever had in this park. It was bigger than Opening Day."

That day, Bonds hit the record-setting homer, and $10 tickets were selling for about $50 in the secondary market. The situation drove home to Stanley something he had heard at an industry conference earlier that year: A vendor told him he was leaving money on the table by setting prices many months before games were played.

This season, to more accurately measure demand for a given game’s tickets, Stanley and his team tested a dynamic pricing algorithm that adjusts prices up to the start of a game. They limited the test to 2,000 of the stadium’s 41,500 seats. After 17 games this year, they saw the following results:
o 20% more tickets were sold in the dynamically-priced sections, compared to last year
o 1.7% more tickets were sold overall, compared to last year

We spoke with Stanley to find out how this pricing system differs from the Giants’ traditional system, how his team manages it, and how it is helping to fill more seats.

-> Pricing systems compared: Tiered system vs. dynamic

The Traditional, Three-Tier System

The majority of the Giants’ tickets fall into three pricing tiers:
o Tier 1: Highest priced, such as Opening Day and late-season games against rivals such as the Oakland As and LA Dodgers
o Tier 2: Weekend games against popular teams, and games determined to be "bigger than most"
o Tier 3: Lowest priced, such as weekday games against poor teams

The prices are set in September for the following season’s games, which start in April. Stanley’s team tries to predict demand for each game from a number of factors, such as the potential draw of specific matchups. But once the season starts, they can’t make adjustments to reflect current conditions.

"If the team is on a five-day winning streak, our ticket is worth more than it would have been than if the team was on a five-day losing streak," says Stanley. "But yet, set the price eight months earlier and we have to live with it."

Dynamic Pricing System

The tested system was based on a pricing algorithm the team created with the help of a statistical analyst. Game prices regularly adjust as several factors in the algorithm fluctuate.

These factors include:
o Starting pitchers
o Weather
o Team performance
o Time of day, day of week of the game
o Special events or promotions
o Previous games’ sales
o Ticket prices in the secondary market (ticket resellers)

Under this system, the prices change regularly. Some tickets for the tested seats have sold for a few dollars more than they normally would have, capturing higher-than-previously-estimated demand for that day’s game.

Other tickets sell for less than the previous price, ensuring that as many profitable tickets are sold as possible despite dampened demand.

The ticket prices do not keep pace with the secondary market of ticket resellers, Stanley says. Resellers have, at times, lifted prices of the upper deck’s normally low-priced seating to $100.

-> Insights into dynamic pricing management

Below are three insights Stanley and his team gained during the trial.

Insight #1. The algorithm is not always right

The algorithm is designed to estimate the best possible price to maximize ticket revenue. To do that, it has to perfectly estimate demand for each game -- which is impossible. There are an infinite number of factors that influence demand.

The best the algorithm can do is estimate -- and it is a work in progress. Stanley and his team have been adding factors and tweaking their weight to "teach" the algorithm how to set the best prices.

"I’ve been doing this for 20 years, and the other person I work with has been with me for 12 or 13 years now. There are certain things that we know will make a price go up or down or stay the same, and the computer doesn’t know that yet."

Two examples:

- During the three-day Memorial Day weekend, the team had to manually set prices higher for the game on Monday, because they knew demand would be higher since many people had the day off. The algorithm wasn’t aware of the holiday.

- Group sales have a tendency to skew the algorithm’s demand estimate. A youth organization could purchase 150 tickets to an otherwise low-demand game, causing the computer to overestimate the general public’s interest and raise prices.

Insight #2. Be cautious

Three ways the team is showing caution with its pricing strategy:

- Do not automate

The system does not automatically update prices, yet. The team is more comfortable with a manual update based on the algorithm’s analysis, especially while the algorithm is still learning how to best gauge demand.

- Keep loyal customers happy

The team tested the two seating areas that are usually among the last to sell. More importantly, no season ticket holders are in this area, which avoids rubbing the Giants’ biggest fans the wrong way.

"I didn’t want season ticket holders paying $12 and having them sit next to someone who came in for $8," Stanley says.

- Test a sample

The team tested a small sample, less than 5% of its seating capacity, to avoid pricing chaos.

"I wanted to just kind of dip our toe in the water. I didn’t want to do the whole ball park. I wasn’t sure that the world was ready for that."

Insight #3. Prepare for a lack of certainty

The team is continuing with the test for the remainder of the season. Stanley is confident that dynamic pricing is helping the ballpark make more money, but he cannot be certain. That’s because ticket prices are just one factor in the team’s revenue stream.

Ticket sales in the tested sections are up 20% -- but revenue is about flat, compared to last year, he says. Overall the park’s ticket sales are up slightly, but the team is also a little better, he says.

However, bringing additional fans into the ballpark, even at a lower ticket price, is likely boosting concession sales, Stanley says. Also, lower ticket prices are likely persuading some people to experience Giants baseball who otherwise would not have.

"I absolutely feel that we’re getting some people in this park who would have not come were they not able to find a bargain," Stanley says. "And I’ll tell you, this park is a heck of a lot more fun when it’s full. It adds to the excitement."

Useful links related to this article:

Creative Samples from the San Francisco Giants’ dynamic pricing test
http://www.marketingsherpa.com/cs/Giants/index.html

Special Report: How to Use Predictive Modeling to Pick Your Best Prospects and Boost ROI Up to 172%
https://www.marketingsherpa.com/barrier.html?ident=3059


AT&T Park
http://mlbnetwork.mlb.com/sf/ballpark/index.jsp

The San Francisco Giants
http://sanfrancisco.giants.mlb.com/index.jsp?c_id=sf