Here's how a data analytics team could be a filmmaker’s best friend.
You made a short film. You spent a lot of time and your hard-earned money on it. Maybe it’s your calling card. Maybe you play some festivals. Maybe you get some words of encouragement. Is that the end of the line? Kat Ascharya, the head of programming at Omeleto, says no. A good short film should not languish in obscurity.
Omeleto releases one short film a day on YouTube. The company is one part-programming, one part data analytics comprised of teammates spread across the world. What started as an experiment during the early Wild West days of YouTube turned into a successful launching point for short films, some landing anywhere from 20,000 to 20 million views.
Ascharya sat down with No Film School to explain why no filmmaker should sit on a short that could be finding a hungry audience in an increasingly mobile age in entertainment.
NFS: The description of Omeleto as “Moneyball for Movies” is very intriguing. What does that mean?
Kat Ascharya: We had a deep background in data and analytics and understood how algorithms attempted to measure audience satisfaction. So it refers to the centrality of data science and analytics to our approach. Filmmakers and studios spend a lot of money in traditional marketing to drive viewers to their films, buying ads or campaigns to get audiences in theaters. You can run paid advertising campaigns if you have the budget but it doesn't guarantee success. Organic reach—real audience engagement—reveals what the audience thinks of your film and what’s working and not working against the sample size. This is, at the core, what platforms try to measure and help them surface the best stuff.
We work as a team with filmmakers to use data and analytics to help their films succeed in the most competitive platforms in the world. There's no "gaming the system," and no amount of data science magic is going to surmount a bad film. But we do devote a lot of time and energy into a/b testing on every minute detail, running countless iterations and then applying insight gleaned from those metrics to tweak and optimize films to, hopefully, reach millions of viewers.
That's how we're also different from a lot of short film channels. We don’t just pick a film and throw it online. We also don't have a niche. Instead, we focus on getting films seen by mainstream audiences at large. Because platforms essentially want to serve content that viewers will love, so they’ll keep coming back for more, the content that’s surfaced has high total watch time, strong engagement, among thousands of other metrics.
"These small adjustments can make a huge difference in the trajectory of a film’s success – the difference between a few thousand or a few million viewers."
NFS: So Omeleto uses a proprietary dataset based on previously published films. Can you elaborate on how this is different than, say, a filmmaker putting in relevant tags or taking out ads and asking every living friend and relative to share, share, share?
Ascharya: I tell all the filmmakers we work with that releasing a film online can't be approached in a traditional film industry “opening weekend” way, where you work towards a set of screening or release dates and drive a huge spike of viewers or awareness to it at the beginning. If you want a significant audience, you don't want to do that. Platforms have smart algorithms that limit reach for videos with huge numbers of clicks but low audience retention, for example.
Instead, you need to focus on the underlying metrics, such as total watch time, audience retention, engagement, etc. It does help to share with your "fans," but they can't just click and leave—they need to show genuine interest and attention and watch a significant portion of the film. So, thoughtful sharing always helps. Watch time and engagement are important, but you can’t "game" those either. Once the sample size of viewers becomes large enough, the metrics always stabilize to the mean.
We can't get into too many details of our process—it is pretty complex and super-geeky—but having worked with thousands of films, we do have a store of data based on published films that we’ve gleaned actionable insights into the elements that affect a film’s performance. Using this, we have an a/b testing process, where we test out all facets of a film prior to launch. We test a lot and run countless iterations! But more importantly, we monitor the analytics after a film is released and make incremental adjustments based on the real-time data and feedback we get that tells us what's working and not working. These small adjustments can make a huge difference in the trajectory of a film’s success—the difference between a few thousand or a few million viewers.
Regarding techniques, like tags... anything that can be easily manipulated tends to be obsolete quite quickly. It's a cat-and-mouse game with platforms and creators— if too many people know that tags work, for example, everyone starts spamming tags, and if it works, poor content surfaces to the top. Then viewers complain about poor recommendations and the platforms change the algorithms. Tech companies have teams of very smart PhDs thinking of ways to surface relevant content. They're constantly updating their algorithms to fight against manipulation. And when they do, those new algorithms can take down huge sites and channels that have over-relied on those techniques to get clicks and views.
We don't rely on hacks and tricks. Instead, what's important to us is testing to understand the underlying metrics, then applying those metrics to glean insight into human behavior and quality storytelling. For instance, how audience retention relates to the opening hook or how a plot twist translates into engagement.
"...any video on any platform, about half of its viewers click off a film within the first minute. "
NFS: Are there aspects of good filmmaking that Omeleto just can’t find success with? As in, is there a home for super intentionally slow videos à la Satantango, or do you have to forget certain styles of storytelling to find success on Omeleto?
Ascharya: Most social platforms are a very large, broad, and general audience, so more traditional or classical forms of storytelling do find a larger, broader audience on them. That's true before online video and I don't see that changing.
I do find the fundamentals of solid storytelling always resonate, no matter what technology or platform is in vogue. People want interesting characters with strong wants; they want real and believable stakes and plausible obstacles. They want a modicum of solid craftsmanship and good performances, and they like an arc that feels like it's taken them on a journey. And that's reflected in the underlying metrics.
That said, I'm always surprised at what resonates. There are always fascinating outliers and exceptions, like where an experimental film about an iPhone murder mystery somehow finds an audience who lives much of their lives on their phones, or when a more avant-garde (but still story-driven) animation starts to surge because its narrative is still relatable underneath the unique visuals. That's perhaps the algorithm working the way it should, finding those viewers at large that would resonate with those films.
Filmmakers do have to realize online is a highly competitive environment—with any video on any platform, about half of its viewers click off a film within the first minute. That's just the nature of the beast, and true across the board, and makes it harder for more abstract, slower-paced work or stories that take a while to get going.
I do think it's important to understand what we do can't make up for poor storytelling, and filmmakers need to be clear and honest with themselves about their work and the audience they had in mind when they first generated the idea or went into production. Filmmakers still need to make great films that are authentic to their voice and vision above all else. That said, once it’s made and distributed, it enters a whole new arena, and if social platforms are a part of that distribution strategy, they’re entering a very competitive landscape. A lot of great films never get seen, because you still need to break through the clutter of content. There's no shortage of videos to watch or things to read or pictures to scroll through. Our hope is to optimize and push great films into the massive content discovery engines that are these platforms and get them the audience that will appreciate them most.
"...we currently limit releases to one film a day."
NFS: What do filmmakers get out of it, and what’s the Omeleto philosophy about the aspect of monetizing and how filmmakers make back?
Ascharya: Filmmakers who work with us get the guidance and expertise of our data scientists and engineers, so they can focus on telling stories, finding collaborators, and getting funding. Most filmmakers I know don't want to keep up with the ins and outs of metrics and analytics. They want to make work or make connections that will help them make work.
That said, we always share a lot of the nitty-gritty in terms of metrics, algorithms and such, especially to assist in filmmakers' receiving real feedback from viewers on a massive scale— insights they might take to hone their craft with. Data and analytics is not information you'd get in a film school, but it's helpful to know in today's fast-moving media landscape, and many filmmakers we work with, being curious artists at heart, find it fascinating and eye-opening.
Two other problems filmmakers tell us they have are: 1.) How to get an audience for their film, and 2.) How to make money to support their work.
We've done pretty well at solving their first problem with data and analytics. We've been exploring how to solve their second problem. A few months ago we started testing revenue share for certain high-performing films. We still release most of our films without ads. We didn't get into this to make money. But for certain films that seem like they'll gain significant momentum, we offer this to filmmakers. It's still a beta test and it's invite-only to a limited few. But we're always trying to figure out how to help filmmakers achieve their dreams, and help them get the widest and most appreciative audience for their work.
" Data and analytics is not information you'd get in a film school, but it's helpful to know in today's fast-moving media landscape..."
NFS: How do filmmakers submit a film? Any tips for what Omeleto is looking for?
Ascharya: They can submit to us via our website. Our site explains more about our approach, our terms, and process. Since our process is rather intensive, and we prefer to give each film the attention and care it deserves, we currently limit releases to one film a day. We also had to implement an Oscar-qualifying festival requirement to help us manage the volume of submissions.
That said, I think it's also good to know that Omeleto isn't an audience of cineastes—the core audience we build from isn’t going to attend film festivals or want to know what lens was used during certain shots. They're really general interest viewers that appreciate a great story or piece of entertainment. For that reason, we don't specialize in genre or audience—we really just want to find compelling storytelling to share, whether it’s drama, family, fantasy, horror, comedy, or some mix in between.
Personally, I especially look for emotional engagement and intelligence in a film, and a certain clarity in the storytelling. This doesn't mean "feel good" endings; audiences know when they're being over-manipulated. But it does mean meaningful stakes for a strongly defined character, and some kind of arc usually. There's a lot of room for creativity and experimentation within that, but that's the core I'm looking for. A great hook at the start goes a long way and a plot twist at the end is fun, but those are no substitutes for genuinely earned emotion. Audiences are hungry for strong emotional experiences and great stories, and I don't ever see that appetite being whetted, no matter what technological avenues open up.
Thank you, Kat!
Interesting in finding out more or even submitting a short to Omeleto? Check them out here.