Village Roadshow uses predictive rostering for its casual workforce

Village Roadshow has embedded predictive analytics to forecast attendance and optimise casual labour costs, following an eight-week development phase that cost them just “two gold class tickets”.

Chief information officer Arul Arogyanathan said the entertainment company used Qlik Predict to build rostering models for its 4200-strong casual workforce.

Speaking at the Gartner Data & Analytics Summit in Sydney last month, Arogyanathan said he approached the project with the business case of reducing the company’s $22 million casual labour cost.

“We are talking about cost optimisation as one of the outcomes that [the] business wants to achieve,” he explained.

“How can I more accurately roster my labour? And, as a result of that, can I achieve five percent ... [or] 10 percent [reduction] in that labour cost?”

He added that before even exploring a technology solution, Village Roadshow had to consider the commercial justification for any investment.

“We are owned by a private equity firm. If you make a $1 investment, you have to get three times [the] return or they won't give you money.”

As such, Arogyanathan was keen to make use of his current technology stack.

Village Roadshow has used Qlik since 2018, initially implementing the business intelligence solution QlikView, followed by its successor, Qlik Sense. It also uses Qlik Cloud Analytics, a platform for data integration and insights.

“We reached out to our [Qlik] account executive and said: “What can we do in the advanced analytics space?,” Arogyanathan said.

This engagement led to Village Roadshow drawing up a three-page document outlining up to 10 of the company’s main challenges and the likely return on investment from solving these.

From the blueprint plan, Village Roadshow conducted a four-week pilot program after which Arogyanathan was ready to present the proposal to Village Roadshow’s board -- which immediately asked how much had been spent so far.

“Two gold class tickets - that's exactly what we spent,” Arogyanathan said. “One for [Qlik’s] account manager and one for [a Qlik] senior engineer.”

The scoping phase resulted in a four-week deployment of Qlik Predict, a no-code platform used to underpin Village Roadshow’s rostering across six theme parks since the end of last year.

“If you want to realise that value, what is the delivery mechanism that we are going to use to prove the value before we make an investment?, Arogyanathan added.

“With the zero investment, we were able to showcase the value.”

Reducing manual intervention

Arogyanathan first joined Village Roadshow as its CIO in 2021 and decided to address both the company’s siloed data and its ability to use it in decision-making.

“There was a big disconnect between where our data was and how we were making the decisions, and the gap was creating [a] cost implication for the business,” he said.

However, its investment in data analytics over the last three years has now positioned it to begin exploring both generative and agentic artificial intelligence.

The company has now created a fully-integrated model that melds its predictive rostering with workforce optimisation, essentially enabling Village Roadshow to autonomously update its staff roster in under two days, based on business needs.

Although the company has a separate model to do just that, “manual intervention” is required to link this to its predictive rostering model.

“What that means is we do the attendance prediction seven days in advance, four nights in advance, monthly in advance, but we have the ability to change our rostering two hours or two days in advance,” Arogyanathan explained.

“How can I make these two models integrate, so that I can take the human element out of it? That's agentic AI for us.”

Arogyanathan noted that his assessment of whether an AI is truly "agentic" hinges on whether it meets three criteria: context, an ability to reset itself and an ability to make a “goal-oriented decision”.

“We did not start with the mindset that we want to build autonomous agents,” he added. “What we were trying to do was move from a reactive automation to a proactive orchestration.”

In addition, Village Roadshow recently adopted Qlik Answers to underpin an internal chatbot known as VR Concierge for its 5000 staff.

Linked to its SharePoint data, VR Concierge allows all staff members to extract information relating to anything from their leave balance or Village Roadshow’s internal policies.

On the customer-facing side, Village Roadshow recently built another chatbot for its cinema business, which is programmed to answer “static, generic” questions relating to film screenings and showings at cinema locations.

The company has since started progressively injecting more data sources into the chatbot, such as its live database of film release dates.

The goal is now to “give the customer as much information as possible”, said Arogyanathan.