How AI Eye-Tracking Tech Could Supercharge Film & TV Marketing

If you don’t know your “gaze path” from your saliency rating, it might be time to sharpen up on attention metrics, especially if you have a movie or series to promote. AI-driven eye-tracking tech is now being used to customize and optimize all manner of content that puts wind in the sails of films and series, from trailers to key art and posters and everything in between.

Deadline’s recent investigation into AI-enabled fake movie trailers caused a stir and YouTube quickly turned off ad revenue for two of the channels monetizing them. Meanwhile, AI and algorithm-based tools are being used by agencies that are creating legit trailers, promos and materials for streamers, studios and broadcasters. One such agency is Once Upon a Time, which has garnered awards for its work on movies including Venom and series including The Crown.

It uses Dragonfly AI, originally developed by academics in the UK and which utilizes a patented algorithm to track what the human eye is drawn to. In the entertainment business those insights can be used to tune up a range of materials, but – and here’s where creatives might breathe a sigh of relief – its real-world use currently comes after the design of the key art.

“It has enabled us to algorithmically unpick the human eye,” explains Toby Richards, Once Upon A Time’s strategy and planning director. “The very premise of Dragonfly is that it’s an algorithm-based way of looking at something visually, we are looking to understand what draws visual attention.”

He sums it up: “What we’re then thinking about is saliency. How does something stand out in a world where everything is trying to stand out?”

What Are You Looking At?

Dragonfly AI isn’t the only product designed to show what people will notice most in a design, but it is the one that Deadline sees up and running on a visit to Once Upon A Time’s east London office.

Looking at the hot and cold spots on any given piece of content allows a gaze path to be created. It’s many steps up from a simple heatmap, which is often just a measure of brightness, and Dragonfly tracks what someone is looking at and in what order.

Many industry professionals are justifiably concerned about the impact of AI-enabled tools on their livelihoods, but for Richards, this is additive, and not an alternative, to human artistry.

“It does not replace creativity, if anything, it empowers it,” he says. “You can be confident in knowing something stood out. But this type of tool does only half the job. There is still a need for whatever is standing out to evoke something, and this tool is not currently doing that, the storytelling.”

A Longer Long Tail

The real-world use for Dragonfly AI, in the context of movies or series, comes after key art has been created. It is about making sure assets that will exist across different platforms or even in different spaces, are tweaked and tuned for each environment. A movie, for example, could live across a theatrical release, before heading to a premium sell-through window, then getting picked up by a streamer, before landing with a broadcaster. Each window and platform has its own individual requirements.

“You think about how something might be applied in lots of different contexts and markets and channels. What this enables you to do is take whatever creativity is there and optimize it.”

For content owners and creators, making sure materials and assets are tailor-made for every environment could lengthen the life of their IP, extending the so-called long tail.

“There’s a sense of, how can you make sure you’re going to get the most of something throughout the whole life of a title, and not just its launch,” Richards says. “That’s where, at the moment, this is coming into its own. A lot of the [uses of Dragonfly] are because content is often ending up on [streaming] platforms, but it may have started out life as a theatrical piece. The creative asset designed for theatrical is not necessarily the best articulation of what’s interesting when it ends up on a platform.”

Human Meets Artificial Intelligence

“The question is not what you look at, but what you see,” wrote American essayist Henry David Thoreau back in 1851. Fast forward to an AI-enabled entertainment biz in 2025 and there are tools that purport to help interpret what is seen. But human intelligence is key to the process.

“I don’t think this is a panacea,” Richards says. “I think it’s an additional, incredibly effective addition to the toolkit. Will there be other additions to toolkit? Yes. Should you push your tools over? Definitely. It’s helping us observe differently, but we’re still doing the observing.”