The Reconnection Gap in Streaming

The problem

People use TV as an escape from busy lives. The problem is that life keeps pulling them out. While they watch, everything in the room competes for their attention. That is cognitive load. Once they stop, everything else in their lives pushes out what they saw. That is cognitive load too. Then they try to come back, and getting back into the story is one more load on top. So the escape becomes work and they do not go back. That last load is the one to minimize. Remove it and people get back into their shows and enjoy them. That has implications for how the living room gets monetized, and it stops people getting the most out of their stories.

Reconnecting to content you were watching before is like meeting a friend you have not seen in a few months or a year. The first thing you do is catch up, what has happened in their lives and why, who is involved, and what it means, before that connection is reestablished. This core principle of connection comes from cognitive science. This is why features that provide context and continuity matter.

The complication is that today, streaming platforms know what you watched, when you stopped, and what you finished. That information powers restarting points, recommendations, and notifications, and it works. What it does not do is connect you back to the story after a long pause, when you have potentially abandoned the content. Not the characters, not the story arc, not what is inside the story. It cannot shape what you watch to the time you have, find the show your whole family will sit through, or connect you to what you see on the screen.

Reconnecting a viewer to the content after a long pause, or after they have abandoned it, means understanding the story, its characters and its arc more deeply than what is done today. It is like unlocking the content and the story from the screen.

What was done

Nine feature ideas were explored, all of them reflecting real pain points in the living room. Every one of them works against cognitive load, so each idea either prevents abandonment, recovers engagement that would otherwise be lost, or grows engagement.

The core themes tested were these five.

Reconnection

Reconnecting to the story, its characters, plot, and why it all matters.

Attention

Staying or re-engaging with the story through interruptions and drifting interest.

Time

Fitting a story into the time you have without losing its essence.

Decisions

Deciding what to watch together when no one can agree.

Curiosity

Getting to what you see on screen.

To understand whether these pain points resonate, the ideas were tested with 100 streaming viewers.

The goal was to understand if the pain points resonate, if the potential feature ideas addressed them, and to identify any undiscovered features in this space. To test this, a 15-question survey was fielded to find out what people actually need in the living room. Every idea was framed the same way, as a television that talks to the viewer about what they are watching, and in some cases answers when the viewer asks. The same would apply to any smart speaker.

What was tested

The nine feature ideas

How it was tested

What was learned

Pain points reported

83 percent report at least one pain point, and 64 percent report two or more. All of it asked before any feature was shown.

The three largest are the same problem at three stages. People lose interest partway through an episode and quit, at 40 percent. They drop shows entirely and never come back to them, at 45 percent. And when they do try to return, they cannot remember what happened, at 60 percent, which is the largest of all seven and a third larger than the next. Four of the seven sit under reconnection, including the two largest.

“Don't Stop Now” has a high pain point and low interest, which could mean the feature is not aligned with the problem it was meant to solve.

“Who Is This” has a low pain point and the highest interest of all nine, which could mean people did not realize they had the problem until they saw what the feature would do, and it resonated.

Both are signals on where to focus. There is something in each of these two areas, and it might not be exactly what people think it is. For a product manager deciding where to spend the next round of work, that is where the next test starts.

The features, ranked

High interest with few picks does not mean an idea is unimportant. It means it is not the highest priority right now.

Follow Me Back had the third highest interest and finished eighth on priority. Fit My Time had the lowest interest and finished fourth on priority.

Three feature ideas rose to the top and are worth taking forward. Features 1, 2 and 3 carry an added benefit, because they are potentially the same solution and one foundation could serve all three. “Fit My Time” is fourth and the first one below the line, so features 3 and 4 would be validated further with additional testing or prototyping.

Recommendations and rationale

These are the top feature ideas that warrant more investigation and refinement for potential product investment. They could be the killer ideas. You talk to your TV and it gives you what you need to reconnect faster. The recommendation comes from the ranking and from the comments people wrote in their own words.

1. “Catch Me Up (Story Aware)” It leads on all three columns, the pain point at 60 percent, interest at 78 percent, and 72 percent of the top three picks. The comments reinforce it. 32 people named catching up or recaps when asked which features they would use, and five more described a catch-up feature without ever having been shown one.
2. “Who Is This” It carries one of the smallest reported pain points at 22 percent, but the highest interest of all nine at 80 percent and 58 percent of the top three picks. The comments explain that gap. 14 people commented that remembering characters is one of the issues they would like solved, which increases the need for this feature above and beyond the 22. This could have indicated the question was not clear for consumers, but the input was direct and overwhelming.
3. “Come Back” It had a good ratio of pain point reported, interest, and selection, at 45, 46 and 37 percent. Nobody decides to quit a show, and 45 percent said they drop shows entirely and never come back to them.

“Fit My Time” is fourth, and the first one below the line. There was interest for this feature, and it does merit further evaluation.

Features 1, 2, and 3 carry an added benefit. They are potentially the same solution, and one foundation could serve all three.

What would change these recommendations

Further exploration and testing. This was a concept and discovery pass. The goal was to identify direction and where further exploration should happen, not to draw final product investment conclusions. These recommendations hold until that exploration and testing say otherwise.

Key learnings

About the pain points

1. The pain points were device agnostic. Pretty much all of the pain points tested were spread across all the different devices. This indicates a gap in the entire category, not on one platform.
2. Reconnection was the most important theme. The top three features were all about consumers reconnecting with the story and the characters in a more meaningful way after long pauses.
3. Many users struggle with reconnection. The largest pain points were about reconnecting with the story after time away. The comments name the same two features. 32 people wrote catching up and 14 wrote identifying characters, in their own words, and five more described a catch-up feature without having been shown one.
4. The reconnection issue has not been solved yet. Solutions that aim to solve it exist today, previously on clip reels and video recaps. Across the board people still reported the pain point and still wanted a solution.
5. People identified other pain points that have not been addressed. Not having time for a full show, losing interest in an episode, and not being able to agree on what to watch. They ranked below reconnection but still showed sizable interest and demand, and today there is no solution in the market that addresses them. That could be an opportunity to validate them further and determine whether there is differentiation there.

About the feature ideas

6. Demand concentrated in three reconnection ideas. The survey results not only validated reconnection but prioritized it. Catch Me Up (Story Aware), a recap that explains what matters and why. Who Is This, asking who a character is and why they are in the scene. And Come Back, being brought back to a show you gave up on. Those three share the same foundation. Build it once and the three features are doors into it.
7. “Fit My Time” resonated with families, younger viewers, and men. 41 percent of households with children picked it against 19 percent without, 43 percent under 35 against 17 percent of 55 and over, and 33 percent of men against 17 percent of women. This could indicate that families have a greater struggle with time management. A larger test would most likely raise this.
8. “Watch Together” resonated with younger viewers. 32 percent under 45 picked it against 15 percent of 45 and over. This could indicate that younger households do more social viewing, which requires keeping two people in sync on the same show.
9. “Come Back” resonated with older viewers. 51 percent of viewers 45 and over picked it against 17 percent under 45. This could indicate that older viewers are more likely to start a show and abandon it.
10. “Follow Me Back” resonated with men. 24 percent of men picked it against 6 percent of women.

What this points to

11. The third feature is a product decision. Catch Me Up (Story Aware) and Who Is This lead in every group. After that the audience splits. Older viewers without children want Come Back. Families and younger viewers want Fit My Time, and they lift Don’t Stop Now and Room Match with it. Time and attention are family problems. Who the product is for decides whether the third feature is Come Back or Fit My Time.
12. There is an affinity between the top three feature ideas. Given the concentration of demand around reconnection, the unlocking move is solving the reconnection problem in a scalable way. Solving it through a platform provides the foundation that makes all of them possible, and the end result is that the investment compounds across features rather than being spent on each one.
13. Some of these features would need to know who is in the room. 48 percent refused every capability when permission was asked cold with no feature attached. From a privacy standpoint users seem hesitant to allow that kind of detection, so it would need to be handled carefully.

About the method

14. Forced choice added a different dimension to the feature prioritization. Interest showed what people wanted, and most features scored well on their own. Having to pick three out of nine showed what they would give up, which helped in understanding what features to prioritize.

What each idea changes

How each feature idea would impact the business

These features either drive more engagement or recover engagement that would otherwise be lost, by enabling the user to connect with what they are watching and with the story. In industry terms, preventing abandonment could raise completion rate, recovering engagement could improve retention and reduce churn, and growing engagement could raise hours viewed.

What this all means

The finding

There is an opportunity to address these use cases and improve the viewer’s experience in the living room. The underlying technology exists today, built for other purposes. Story graphs, also called narrative graphs, are used to check a storyline for continuity, to carry a show’s canon into a game, and are offered to AI developers as a way for a model to play a character. Those graphs can be extended. The meaningful work is taking them and making them usable in the context of the living room. That is not retrieval of information. It is attaching the relationships between characters to a timestamp, because what is true between two people changes as the story moves, and the window the viewer is allowed to see moves with them. AI then extracts what matters at that point and conveys it to the viewer based on where they are, rather than using the graphs as a base to create something, to search something, or to play a character.

This concept test set out to understand how people reconnect with content, given the cognitive load it takes, and whether they want a television they can talk to about what they are watching. Platforms are trying to solve reconnection today. Most play the same clip recap for everyone, and the few that generate a recap summarize what happened rather than who wants what and why, and none answers from where you are in the story. There can be a broader, more intelligent solution that provides greater context faster and lets consumers reconnect with their content at any point in time, by asking. That benefits the business, and one way to build it is the story graph described below.

Three feature ideas rose to the top, and all three attack the same problem, the cognitive load of reconnecting with a story. That concentration is the finding, and it says this area merits future investment.

What is missing today

The solutions that exist today are not solving it. Transcription, scene description, and information about the actor tell you what is on the screen, not what it means. What people want is that this person is connected to this other person, and they matter in this scene because of what they did earlier. That is context, and it is what is missing. That gap is the opportunity.

One hypothesis is that what is missing is story awareness. A recap today replays the same fragments for everyone, or summarizes what was said rather than what it meant. A story aware recap knows the motives, the stakes, and what each character knows, so it gives you back the story rather than the facts. The same holds for a character. Pause and search today and you get a name, a face, and a page written about the actor. A story aware answer knows the character history, their relationships, and what they want, and it knows how far you have watched.

One way to solve it, a story graph

One potential way of solving this is a story graph. Imagine asking a friend who has watched every episode to catch you up. They tell you what happened while you were away, who did what to whom, who is who, and what it means. They know each character, what that character means to the story, and what the characters mean to each other. In Mexico this is called “chisme”, and it brings someone up to speed on whatever is happening in the community in just a few minutes.

A story graph is that knowledge, written down. Cognitive science says our minds follow a story by tracking time, place, cause, what each character wants, and who is doing what, and the graph carries each of those.

It holds everything that happens in the story, who is involved and why, with each piece stamped with when it happens. Ask a question at any point in the story and the answer comes from that knowledge, up to that point and no further. This is the kind of information it would carry.

Because every element carries a position, a question can be answered at any point in a story without revealing what the viewer has not reached.

How it could be built

Today, story graphs, also called narrative graphs, are already in use. However, they serve a very different and very specific purpose. Writers use them to check a storyline for continuity. Game studios use them to carry and reproduce worlds in a game. AI developers use them to build specific characters.

Much of the information in these graphs includes the characters and the events, but it does not include how relationships change, or timelines. That is the gap, because in the living room the one thing that matters is the timeline. The plot develops and evolves, and what is true between two characters changes as the story unfolds.

When viewers want this information, it is all based on the point in time where they have viewed the content and what is happening on the screen. It is not about the entire story. It is about what is happening in the story at a particular time, scene or character.

That same approach already used in the industry can help solve these living room pain points, but it must be extended to merge what is happening on the screen, the time, and a story graph that carries relationships.

How this could work in practice. A living room device, the TV or a companion device, understands what is happening on the screen. It knows the title, where the viewer is in it, and what is on the frame. It sends that back to a server that holds the story graph. The server takes the timestamp, cuts the graph at that point, and returns what is true up to it. Who the characters are, the relationship between them, what has happened, and what each of them wants, and nothing past that point.

A majority of the existing infrastructure today could be used to develop and extend these solutions. Video analysis and content delivery already run at scale. What changes is an additional payload, not in the video itself but alongside it as metadata and the story understanding.

The TV sends where the viewer is and what they need. The cloud reads the story understanding and returns markers, which go back to the television or out to a smart speaker. None of it is video, and the video that reaches the television is the same video it would have delivered regardless.

Where it reaches

The story graph can be extended to enable other features. Seven of the nine feature ideas draw on this same understanding of the content.

1. Fit My Time needs to know which scenes carry the plot.
2. Don’t Stop Now needs to know where the next turn is.
3. Watch Together needs the difference between two positions.
4. Follow Me Back needs the timing of a moment worth coming back for.
5. Room Match can use it to know what is actually in a show, how scary, how funny, how intense. That is more than genre. It is looking inside the content’s overall storyline.

One understanding of the content helps all use cases. Solving these top pain points is not about building a single feature. It is about building a foundation, an understanding of the content, that makes the rest possible, which means the investment compounds across features instead of being spent nine times.

Method and full results

Back to Portfolio