The Metrics That Built Social Media May No Longer Matter Most

If someone spends 60 minutes scrolling through 100 videos, most social platforms have plenty of reasons to call that a successful session. If someone spends those same 60 minutes with one person – talking, reacting, requesting a song, coming back to a conversation – the numbers may look less impressive. But which experience was actually more social?

I think about this distinction a lot because I operate a social livestreaming platform, and our industry has become exceptionally good at measuring consumption. We know how long people stay, how much they watch, what makes them stop scrolling and what brings them back. These metrics helped build some of the most successful products in history.

The problem starts when we assume they also measure connection. Someone can spend an hour surrounded by people on a screen without interacting with a single one. We’ve all done it. You open an app to reply to a message, somehow end up watching a stranger reorganize her kitchen, learn more than you intended about someone else’s divorce, see three opinions about AI and emerge 45 minutes later having communicated with absolutely nobody. That’s engagement. It just isn’t necessarily social.

Many social platforms began primarily with people we knew or deliberately chose to follow. Then recommendation systems became good enough that knowing someone was no longer necessary. The feed could simply find something we’d probably like.

It was an extraordinary product breakthrough. It also changed the behavior of social media. We increasingly stopped opening apps to see what our people were doing and started opening them to see what the algorithm had found for us.

Now, generative AI is removing another constraint. The internet already has more content than anyone could consume; producing the next image, video, song or post is becoming dramatically easier. There will always be something interesting waiting for us.

So the more interesting question for the next generation of social products may not be what can we show you next? It may be: Who is there when you arrive?

The same 60 minutes can create something completely different

Imagine one person spends 60 minutes watching dozens of videos perfectly selected for his interests. He laughs, likes two and closes the app.

Another spends 60 minutes watching a musician perform live. She requests a song. The musician recognizes her from earlier in the week and jokes that she’s asking for the same song again. Other regulars join in. Someone remembers she had a job interview and asks how it went.

Both users gave the platform 60 minutes. The first experience produced consumption. The second produced history.

Human relationships accumulate context. You remember what happened last time. You develop inside jokes. You recognize names. You notice when someone who is usually there isn’t. And unlike content, that history can’t simply be produced once and distributed to ten million people. It requires participation. The difference is that you mattered to what happened next.

Some good social behavior looks bad on a traditional dashboard

This is where the question becomes interesting for product leaders. A user who stops discovering new people because she keeps returning to the same five may look less exploratory. Someone who spends an hour with one creator instead of watching 30 may consume less content. A small group that spends months talking to one another will never produce the reach of a viral video. But if the product’s purpose is social connection, those might be some of its strongest outcomes.

That doesn’t make views, watch time or daily active users bad metrics. They answer important questions. They just don’t answer every question. Instead of inventing another universal KPI for “connection,” leaders should ask something more useful: What would people do differently if they genuinely valued the people they encountered here?

Maybe they return to the same people. Maybe interaction becomes reciprocal. Maybe they remember previous conversations. Maybe someone opens the app because a particular person is there, rather than because an algorithm has promised an endless supply of entertainment.

The answer should be different for a dating app, professional network, multiplayer game or livestreaming platform. Their definitions of a successful human interaction shouldn’t be identical. And AI makes that distinction more urgent. It will make content more abundant, more personalized and probably more compelling. The marginal cost of giving us one more thing to watch, read or listen to is heading toward zero. That makes attention easier to fill, but it doesn’t make relationships easier to create. Technology will become even better at ensuring we’re never bored. But eliminating boredom isn’t the same as creating connection.

There will always be another video to watch, another post to read and another perfectly optimized thing waiting one swipe away. The harder product challenge is giving someone a reason not to swipe. Because sometimes the best sign that a social product is working isn’t that people want to see what comes next, it’s that they’ve found someone worth staying for.

If someone spends 60 minutes scrolling through 100 videos, most social platforms have plenty of reasons to call that a successful session. If someone spends those same 60 minutes with one person – talking, reacting, requesting a song, coming back to a conversation – the numbers may look less impressive. But which experience was actually more social?

I think about this distinction a lot because I operate a social livestreaming platform, and our industry has become exceptionally good at measuring consumption. We know how long people stay, how much they watch, what makes them stop scrolling and what brings them back. These metrics helped build some of the most successful products in history.

The problem starts when we assume they also measure connection. Someone can spend an hour surrounded by people on a screen without interacting with a single one. We’ve all done it. You open an app to reply to a message, somehow end up watching a stranger reorganize her kitchen, learn more than you intended about someone else’s divorce, see three opinions about AI and emerge 45 minutes later having communicated with absolutely nobody. That’s engagement. It just isn’t necessarily social.