Machine Learning Times
Machine Learning Times
EXCLUSIVE HIGHLIGHTS
Watch 3 Videos from Coursera’s New “Machine Learning for Everyone”
  I’m pleased to announce that, after a successful...
97 Things About Ethics Everyone In Data Science Should Know
 Every now and then an opportunity comes along that...
Machine Learning is Transforming Modern Healthcare
 The pandemic has propelled the adoption of innovation and...
It’s a Bird, It’s a Plane, It’s a Classified Flying Object
 How Computer Vision Is Used To Classify Objects. Featuring...
SHARE THIS:

6 months ago
Wise Practitioner – Predictive Analytics Interview Series: Reeto Mookerjee at Fandango

 

By: Eric Siegel, Founder, Predictive Analytics World for Business

In anticipation of his upcoming conference presentation at Predictive Analytics World for Business Las Vegas, May 31-June 4, 2020, we asked Reeto Mookerjee, Vice President, Data, Analytics and Business Intelligence at Fandango, a few questions about their deployment of predictive analytics. Catch a glimpse of his presentation, Fandango 360: Data Driven Movie Marketing and Content Recommendations Platform, and see what’s in store at the PAW Business conference in Las Vegas.

Q: In your work with predictive analytics, what behavior or outcome do your models predict?

A: There are three classes of problems, broadly defined.

On one hand we predict the propensity of each and every audience in Fandango’s digital network to watch a particular movie in the theater, as well as, the propensity that whether they will rent or buy a movie at home (either a new release or a catalog movie). For the theatrical and new release home entertainment, we need to come up with these predictions way before we actually observe any interaction (e.g. watching trailers etc.) of the audience with the particular movie. We have turned this into a data platform/product (Fandango360) for our studio partners, their agencies as well as non-studio partners.

The second class of problems are around forecasting science. We predict the opening weekend estimates of movies, typically 1-3 weeks prior to the movie opens. Then we predict how many tickets Fandango will sell in the next hour, this week, this month, and the next year (for long range planning), how many copies of titles we will sell on a given week for home entertainment, and how many of these will be rentals vs. purchases etc. On the traffic side, for our ad sales business, we predict overall traffic volumes by properties, by future title specific pages (e.g. how many visitors we will have on the Wonder Woman 1984’s movie overview page on the week of its release in early June this year etc.). For Fandango’s digital and retail gift card portfolio, we forecast activation (sales out) as well as redemption across channels.

Lastly, on the classical marketing analytics front, there are predictive customer lifetime value (LTV) models, as well as (content aware) churn propensity for our theatrical and home entertainment customers.

Q: How does predictive analytics deliver value at your organization – what is one specific way in which it actively drives decisions or operations?

A: With Fandango360, most of the major studios are using the propensity scores to target the audience for their movie marketing across paid social (Facebook, Instagram, YouTube ads), video display advertising, and some cases in their linear TV buys (optimized linear plans). These are used for over 100 movie release campaigns across all size and genre, consistently these audiences perform better (on a cost per basis) on average 50%+ than the next best alternative in the respective social platforms. Studios and their agencies are able to market to the right audience for their movie with the right set of creatives. For example, if you are a fan of Amanda Seyfried, then when Universal Picture markets you the Mamma Mia Here We go Again!, you should see the creative where she is in the front and center via creative optimization … kind of that holy grail of marketing.

Q: Can you describe a quantitative result, such as the predictive lift of your model or the ROI of an analytics initiative?

A: For the studio campaigns, we observe if the general awareness of the movie is there (i.e. if the studio is spending on TV and out of home etc.), then the lift(%) against a holdout/control group (exposed to placebo ads) typically is highest from the mid-propensity score ranges, say the score range from 50-80 (on a 0-100 scale). Furthermore, increasing frequency of ad exposures to the super fans (80+) tend to have a diminishing return. On the other hand, for platform releases or movies with low awareness, the lift from the super fans (80+) can be substantial.

In addition to the lift against a holdout group, studios compare efficacy of their media spend on a cost per basis at the two points in the purchase funnel of Fandango — mid funnel where people look at the seat maps of their favorite theater (and many of them bail and buy at the box office) as well as, the down funnel (those who buy tickets on Fandango). When you compare cost par at these two points, and efficacy of media power by Fandango360 vs. other business as usual targeting methods (e.g. Facebook interest targeting or on YouTube affinity audience), routinely we see high double digit cost per efficiency across the funnels.

Q: What surprising discovery or insight have you unearthed in your data?

A: Fandango is my first foray into media & entertainment industry. I was pleasantly surprised even highly creative product like movies (where no 2 ‘SKUs’ are exactly similar or substitute), how well targeting works. At a macro level, if you consider Avengers End Game, the biggest blockbuster of all times,  it is estimated about 32M people watched the movie in the theater during the opening window, which is less than 20% of movie going audience in North America. So you can imagine for a small-medium title, the audience that ultimate see the movie in the theater is only a small subset of the pent up demand that happens via trailer views and other social engagements. Higher engagement in terms of trailer views and view completes can not always be a reliable predictor of financial success of the movie. That’s where first party movie goer data like Fandango and the exhibitors fill the void.

Q: Sneak preview: Please tell us a take-away that you will provide during your talk at Predictive Analytics World.

A: I hope to share our chronological journey with Fandango360 — this is something that started as a side project of ours (analytics and performance marketing teams), and how that evolved as one of the key company initiatives. Our product decisions, engineering and analytical challenges, trade-off etc. Finally, what’s next on this front…

—————————–

Don’t miss Reeto’s presentation, Fandango 360: Data Driven Movie Marketing and Content Recommendations Platform, at PAW Business on Tuesday, June 2, 2020 from 2:40 to 3:00 PM. Click here to register for attendance.

By: Eric Siegel, Founder, Predictive Analytics World for Business

To view this content
Login OR subscribe for free

Already receive the Machine Learning Times emails?
The Machine Learning Times now requires legacy email subscribers to upgrade their subscription - one time only - in order to attain a password-protected login and gain complete access.

Click here to complete this one-time subscription upgrade

Existing Users Log In
   
New User Registration
*Required field

Comments are closed.

Pin It on Pinterest

Share This