Demystifying Propensity: how the right audiences empower movie marketing
Movio have been leveraging AI and machine learning to help exhibitors and studios for nearly a decade. In fact, there are elements of Movio Cinema EQ that movie marketers have been engaging with for years that they may not realise are built on a type of AI.
Since 2016, one of Movio’s primary applications of AI from our data science team has been the Propensity Algorithm. Much like the algorithms of Netflix or YouTube surface suggested content for viewers to watch, the propensity algorithm provides movie marketers with recommended audiences to target, by predicting the likelihood of each moviegoer’s interest in a given upcoming title.
Demystifying Propensity: what it is and how it works
Movio’s propensity algorithm is a recommendation system. By providing movie marketers with recommended audiences of moviegoers from their loyalty base who have a high likelihood (aka high propensity) to be interested in seeing a given movie, the propensity algorithm enables exhibitors to ensure their marketing efforts reach the most effective audience while tailoring to their interests.
But how exactly does the propensity algorithm find these ideal audiences to recommend?
The key to finding the best audiences to market to comes from analysis of moviegoers’ previous behaviour. Past behaviours such as which movies someone attended, how frequently they visit the cinema, and the genres they enjoy, all feed into the propensity algorithm’s ability to determine how likely that moviegoer is to be interested in an upcoming title.
Let’s take, for example, two recent horror titles to compare: Terrifier 3 and Speak No Evil. Both sit under the umbrella genre of horror, and have similar audience demographics, but there’s a vast difference between a gory slasher and a psychological thriller. Marketing more horror movies to attendees of both titles would make sense, but what about crossover opportunities with other genres?
This is where the propensity algorithm looks not only at similar movies, but at similar audiences. Data reveals more trends of the crossovers between interest in movies of similar genres; data suggests a correlation between R-rated superhero movies like Deadpool & Wolverine or Venom: The Last Dance and gross-out comedies, for example. This year particularly highlighted this with titles like Reagan, a movie that was a political drama on the surface, but in terms of audience, was more akin to faith-based titles.
Because the propensity algorithm is able to find these crossover opportunities of interest, movie marketers are then able to take recommended audiences and market to them in order to influence attendance, even among moviegoers who marketers may not have targeted previously based on traditional comparisons alone.
How propensity differs from forecasting
Recommended audiences represent a moviegoing audience with a high likelihood to be interested in a movie. In essence, a measure of moviegoers’ taste. As opposed to an audience built directly on pre-sales, interest is distinct from intent, and interest alone doesn’t automatically translate to attendance.
While in a perfect world, being interested in a movie would always mean one would visit the cinema to see it, what gets in the way is life. A moviegoer may be extremely interested in seeing the new Sonic the Hedgehog 3 as soon as it comes out, but if they fall sick the week it debuts, have a wedding to attend, or a tight budget that week . . . Any number of circumstances for a moviegoer can change what they’ll do.
What the Propensity algorithm’s recommended audiences offers instead, is the audience which has the greatest opportunity to be converted by the right movie marketing. While it isn’t a forecast, it allows movie marketers to find the best audiences to focus their efforts on to try to boost attendance.
Ongoing innovation paves the way
We are constantly refining and continuing to improve the ways we use machine learning and AI in the data science field. Movio have been working with AI—and Movio users engaging with AI—for years, and there are more innovations on the horizon to put even greater power in your hands.