Streaming Services Are Building Sampling Engines

Documentary+, the documentary streaming service operated by FAST channel studio acTVe, has added a vertical, swipeable video feed, giving viewers a faster way to encounter films through short clips. acTVe CEO Geoff Clark told me the service already has about 100 clips live, with organic engagement arriving before the company has meaningfully promoted the feature. Documentary+ is also adding direct links from each clip to the full film, creating a measurable path from curiosity to consumption.

TikTok trained hundreds of millions of people to swipe through video. Streaming services are now applying that interaction model to catalog discovery.

Documentary+ is building a sampling engine for its long-form library.

Tubi, Netflix, Prime Video, Paramount+, Disney+, Peacock, and HBO Max have launched, tested, or announced variations of the same product idea: show viewers a compelling moment, reduce the effort required to choose, and provide an immediate route into the full title.

The Swipe Now Carries the First Viewing Decision

Streaming services have traditionally asked viewers to choose through artwork, descriptions, rankings, genres, and recommendation rows. A strong sample gives viewers an immediate sense of a title’s characters, pacing, visual language, and emotional stakes. The service can then convert that response into a full stream, a watchlist addition, or another recommendation.

Tubi made that conversion path explicit when it launched Scenes in 2024. The personalized mobile feed lets viewers swipe through clips, like them, save the corresponding titles, or move directly into the full movie or episode. Tubi designed the feature to connect short-form mobile viewing with long-form consumption across devices.

Netflix’s Clips feed follows a similar structure. Users scroll through personalized excerpts and can add the associated title to My List, share it, or start watching. Prime Video’s Clips feature also lets viewers move into titles available for purchase or rental. HBO Max is testing a personalized Shorts feed built around scenes, trailers, and direct paths into the full title

The swipe now functions as a low-friction audition. The service earns a few seconds of attention and uses those seconds to sell a much larger time commitment.

Every Clip Has a Downstream Revenue Job

The economics of a sampling engine extend across the full viewing funnel.

Documentary+ currently views the feed primarily as a discovery tool. Clark said the company’s objective is to introduce people to documentaries and move them toward the full films. The company is prioritizing catalog engagement and full-title conversion during the feature’s early stages.

That approach aligns the product with the larger economics of streaming. A clip can generate additional value through the ad impressions, session time, retention, and audience data created when a viewer starts a 90-minute documentary.

Conversion should become the central measurement framework. Streaming execs should track how often a clip impression produces a save, a title-page visit, a full-title start, a completed film, or a return session.

Paramount+ has framed its own short-form initiative around visit frequency and deeper engagement. In an internal email, product and design chief Dan Reich wrote that the company was trying to jump-start efforts to get 1 million clips into its short-form experience as quickly as possible, showing how rapidly a discovery feature can become a major content-operations mandate.

A productive sampling feed helps the existing catalog generate additional viewing. It can create incremental starts for library titles, improve the yield from licensed content, and expose programming that would otherwise remain buried deep in the interface.

Existing Promo Assets Are Becoming Product Infrastructure

The most revealing line in my conversation with Clark was also the simplest: “We’ve already made the clips.”

Documentary+ had created much of the material for social distribution. Adding the feed brought those assets inside its owned product, where the company can control the user journey and measure the behavior that follows each view.

Streaming services already spend heavily creating trailers, social cutdowns, promotional excerpts, interviews, behind-the-scenes footage, and episodic highlights. Those assets can now support discovery within the streaming experience itself.

An in-app feed turns promotional inventory into product infrastructure.

The operating requirements include ingesting clips, clearing rights, verticalizing footage, attaching accurate metadata, personalizing distribution, and connecting every clip to an action. The workflow can generate meaningful catalog leverage with a relatively modest incremental investment.

HBO Max provides a preview of how scaled services may manage that process. Its system uses an internal AI tool to analyze scene-level metadata and suggest potentially compelling moments. Human editors then select the clips that best represent each title and format them for vertical viewing.

AI can lower the operational cost of identifying candidate scenes. Editorial judgment determines whether each clip creates an accurate and persuasive promise. A sample that overpromises can generate weak completion and teach the recommendation system the wrong lesson.

The strongest clip captures attention and accurately sells the experience that follows.

Sampling Turns Discovery Into a Measurable Funnel

An in-product sampling engine connects clip engagement directly to saves, title starts, completion, and return behavior.

Documentary+’s planned click-through feature will let the company begin linking individual clips to full-film behavior. Over time, that data can answer specific programming and product questions:

Which scenes generate the most title starts? Which subjects drive saves? Which clips lead to completed films? Do interviews perform differently from archival footage? Does a recognizable personality generate stronger conversion than a clip built around the documentary’s central question?

Those findings can influence homepage artwork, trailer editing, audience segmentation, acquisition strategy, social campaigns, and the packaging of future releases.

The clip becomes behavioral metadata.

Clip-level engagement adds the moments, ideas, faces, and storytelling styles that prompt action to the viewer profile.

Netflix is already personalizing its Clips feed around titles it expects an individual member to enjoy. Tubi’s system adjusts recommendations as viewers like and save clips. Prime Video refreshes its selection using viewing history.

The next stage will connect those interactions directly to programming economics. An ad-supported streaming service can evaluate incremental monetized hours. A subscription service can measure retention, viewing frequency, and utilization of titles that already carry sunk production and licensing costs.

A Single Title Can Support Dozens of Audience Entry Points

Sampling gives every title dozens of personalized audience entry points.

A documentary might attract one viewer through its subject, another through its filmmaking, and another through a single emotional moment. A scripted series can generate clips around humor, romance, conflict, talent, spectacle, or cultural relevance. Each sample can target a distinct audience response from the same underlying program.

The feed becomes a continuously personalized trailer channel.

It also creates new value across the catalog. A streaming service with thousands of titles owns thousands of potential clip libraries, each containing moments that can be recirculated as audience interests shift.

Older programming can regain visibility when a scene connects with a current event, a cultural conversation, an actor’s new project, or an emerging viewing pattern.

Netflix’s first experiment with a vertical comedy feed, Fast Laughs, was retired after roughly two years. Its performance showed that familiar interaction mechanics alone won’t sustain usage. Sampling engines need strong personalization, meaningful actions, and a clear path into the long-form catalog.

The current wave increasingly connects every sample to a specific next step: watch, save, rent, buy, or explore. These feeds are becoming conversion products with defined economic outcomes.

Documentary+ Can Build a Faster Learning Loop

Documentary+ offers a useful case study because its scale supports rapid experimentation.

A service with about 100 clips can monitor behavior, identify patterns, adjust editorial choices, and establish conversion benchmarks from a focused initial library.

Clark’s decision was guided partly by Tubi’s experience and a straightforward question: why not test it?

The collapsing cost of experimentation changes streaming product strategy. Existing assets can support a focused launch, and real audience behavior can guide each subsequent investment.

For Documentary+, the first question is whether clips move viewers into full documentaries. Once the click-through mechanism is live, the company can evaluate that hypothesis using its own audience data.

A focused documentary service should also have a strong understanding of which moments communicate credibility, urgency, access, and emotional stakes to documentary audiences. That editorial expertise can make a relatively small clip library highly productive.

Documentary+’s vertical-native interview series, Picture Locked, gives the service another source of inventory built specifically for the format. It also provides a way to introduce filmmakers, subjects, and stories before directing viewers toward the larger catalog.

The feed can become a rapid testing environment for content packaging, programming themes, and audience interests.

The Streaming Wars Take

Streaming services are building sampling engines because catalog abundance has made selection one of the industry’s central product challenges.

Execs should treat the vertical feed as a conversion layer within the streaming experience. Clip views and time spent can help diagnose performance. The business case rests on saves, starts, completed titles, monetized hours, repeat sessions, and retention.

That requires coordination across product, editorial, marketing, data, rights, and content operations. It also requires a clear strategy for every piece of inventory.

Each clip should have a defined job. Some should generate immediate starts. Others should build a watchlist, introduce an unfamiliar title, revive library programming, or teach the recommendation system something new about the viewer.

Documentary+ has entered this shift with a relatively modest clip library and a clear conversion goal. Netflix, Tubi, Prime Video, Paramount+, and HBO Max are applying the same model at greater scale.

The services that build the strongest sampling engines will generate more viewing from the catalogs they already own.