Composing a suite of statistics

AUTHOR: Kirsty Harris   DATE: 30.03.04   ISSUE 1, 2004
Professor Simon Sheather puts statistics into play for the Sydney Symphony.

The Sydney Symphony is not spared the age-old problems that all Australian arts organisations contend with: how to best resolve the tension between artistic endeavour and commercial success. While the Symphony juggles this issue better than most, it is, nevertheless, eager for tools that will make it easier to balance these often opposing objectives.

This is where the AGSM’s professor Simon Sheather steps in. A recent project he conducted for the Symphony (which is the AGSM’s artist-in-residence) shows just how statistical modelling can guide effective decision-making and business planning for maximising revenue.

Sheather uses his statistical modelling to take some of the variability out of business forecasting.

Sheather describes the Symphony’s revenue as a function of ticket price multiplied by occupancy (or tickets sold). “If we can accurately measure the main drivers of the business then we can predict the expected revenue for any individual concert,” he says.

{“The aim of these models is to indicate how the orchestra can maximise future revenue within the boundaries of its artistic requirements.” }

Sheather used ticketing data for concerts performed at the Sydney Opera House during 2001 and 2002 to develop his statistical models. “The aim of these models is to indicate how the orchestra can maximise future revenue within the boundaries of its artistic requirements.”

The models account for the variation in ticket prices across all ticket types, including reserved seating, concert series, subscriber discounts and special offers. Other variables the modelling had to incorporate include: the main composer of each program; whether a program features one or more composers; whether the music is considered well-known; if a program features a famous soloist; the year, month, day and time of the performance; the number of performances; the number of subscription tickets sold; and the number of complimentary tickets provided.

“By analysing these variables we can discover what determines the average ticket price. This discovery allows the Symphony to more accurately forecast its expected average ticket price and, thereby, have a more reliable budget for each program,” says Sheather.

Finding key factors
The models he developed contain all the potential predictor variables for which he received data. This way he could estimate the effects of each variable, having adjusted for the effects of all other variables.

His models revealed a number of key factors that affect the Symphony’s average ticket price. For example, February is the most positive month and November the most negative – a discovery that makes it easier for the Symphony to identify the causes of such variations in price and revenue, and to adjust decision-making accordingly.

Tim Calnin, the Symphony’s former artistic administrator, explains that one of the subscriber benefits is the right to swap concert tickets at no extra cost. “Often by November, subscribers who buy tickets at a discount realise they must use up their tickets quickly or risk losing them,” he says.

“This results in a greater number of discounted tickets appearing in November, which affects the month’s revenue.”

The process also showed that customers are prepared to pay an average of $10 more for a famous soloist. With this kind of information, the Symphony is better placed to decide whether the estimated increase per ticket will cover the extra cost of hiring a famous soloist. However, it also has to factor in artistic purpose.

“Some soloists charge so much that the expense cannot be recouped because the Opera House just doesn’t seat enough people,” says Calnin. In such cases the Symphony has to decide whether the soloist’s cost will be justified by sufficient additional benefits, such as publicity, prestige and musician development.

Sheather’s pricing data included some interesting information but could not give a complete picture unless it was coupled with an analysis of occupancy. His occupancy modelling revealed that the most popular time to go to a concert (perhaps unsurprisingly) was Saturday at 8.00pm. “I then compared the positive effects on ticket price with positive effects on occupancy to show which times, evenings and months delivered higher revenues,” explains Sheather.

More Informed decisions
The next step is to test his models’ estimates against actual data. “We need to test the models by using them to estimate the average ticket price and occupancy for each performance in 2003 – that is, estimate the revenue and then compare the estimates with the actual data,” he says.

“Once the models have been adequately tested they can be used to estimate the revenue for each performance in 2004.”

The Symphony will be able to use the data for more effective programming and planning. For example, if the models predict that a program might struggle to cover costs, management could choose to market the program more aggressively. It might also use the data to offer more concerts in the most popular timeslots to meet market demand more consistently.

Sheather hopes his work will give the Symphony’s management a valuable tool for making more informed decisions about artistic and market requirements that will fulfil cultural and stakeholder expectations as well as increase revenues.