An analytics team sat in a room with a spreadsheet at one point during the production of a recent celebrity documentary, the kind that begins streaming on Friday night and takes over the conversation by Sunday. The spreadsheet included information about how viewers interacted with similar content. where people stopped. where they rewound. which instances prompted them to disclose. which scenes caused the drop-off. Decisions about what went into the trailer were based on that data. Perhaps regarding the film’s production. This isn’t conjecture. The individuals who commission and create celebrity films for Netflix, Max, and other rivals are aware of this approach to documentary programming.
The format of celebrity documentaries has been around long enough to develop its own genre norms. The talking heads, the archive video, and the silent moment when the person clearly becomes upset over something that happened years ago. What has changed is that data, rather than just editorial intuition, now plays a bigger role in determining which archive footage leads, which talking heads appear first, and which emotional moment falls in the third act. Before the cameras roll and before the cut is locked, commissioning decisions and editing decisions are influenced by viewer engagement patterns from earlier documentaries about similar figures, in similar emotional registers, and targeted at similar demographics.
If you’re paying attention, this is most noticeable in the trailer. It’s not necessary for a documentary about a musician whose career included a public breakdown, a return, and an artistic reinvention to begin with the breakdown. However, it typically does. Viewer data consistently demonstrates that emotionally charged content in the first fifteen seconds of a trailer drives the click-through and watch rates that a platform needs to justify its investment, not because it’s the most visually appealing part of the movie, which it may or may not be. The most startling moment is revealed first. It was recognized by the algorithm. It was placed by the editor.
An additional layer is added by the audience targeting layer. Depending on their viewing history, different subscriber segments may see different posters and preview clips for the same documentary. After watching three sports documentaries lately, the competitive career story is emphasized in one of them. After watching three relationship-drama reality shows, one might see that the personal life narrative is emphasized. Nothing has changed in the documentary. However, depending on what the platform’s analytics indicates will turn each viewer segment into a watch, different audiences will see different versions of the documentary—the frame through which they first experience this person’s tale.
The most subtle and potentially most significant impact on the story itself is trend matching. The social media landscape is fast-paced, and subjects whose stories align with popular discussions receive different levels of platform support and promotional investment than subjects whose stories don’t. Documentary productions, on the other hand, have lengthy timelines. A celebrity whose circumstances relate to a current discussion on Twitter or TikTok is a more profitable subject for a documentary than someone equally fascinating whose narrative doesn’t fit into the current discourse. This leads to selection pressure over time. Both what is intriguing and what is algorithmically timely are reflected in the tales that are created and the perspectives that those stories adopt.
In this process, the subjects of these documentaries are not totally passive. Celebrities who have a lot of power, especially those who land big streaming documentary deals, are more likely to negotiate creative control provisions that restrict the extent to which the platform can impose data-driven editorial modifications. The industry’s talent side is becoming more conscious of what the data side can accomplish, as evidenced by the fact that these provisions are being sought and negotiated.