The algorithm doesn’t know whether your work is good. It knows what people do with it.
You can produce the most beautifully crafted film in your category. Perfect lighting, impeccable script, professional everything. If people watch it politely and move on, the platform learns something about it. Meanwhile, a video shot on a phone in a parking lot travels everywhere, because people send it to a friend, argue about it in the comments, imitate it, save it, and make their own version.
Craft is invisible to the algorithm. Consequence is the only thing it can see.
That distinction has become one of the most important facts in marketing, and most brands are still on the wrong side of it.
What the Platforms Actually Measure
There’s no universal algorithm. Every platform optimizes for different things, and every surface inside a platform ranks differently. But the drift is consistent, and the platforms have been unusually candid about it.
In January 2025, Instagram head Adam Mosseri told creators plainly: “The top three signals that matter most for ranking are watch time, likes and sends. So when looking at your insights, pay close attention to average watch time, likes per reach, and sends per reach.” He’d been pointing at that last one for months, describing sends per reach as one of the most important signals the platform uses.
That last metric is worth reading twice. Not reach. Not impressions. The ratio between the people who saw the work and the people who forwarded it to one specific human being.
Read it as a creative brief, and it asks a single question: Did this give somebody a reason to put their own name on it in a private message?
TikTok describes its recommender systems the same way in its own documentation. Interactions become “signals that help the recommender systems predict content you’re more likely to be interested in,” and every candidate video gets ranked on predicted interest rather than on any assessment of the video itself. YouTube has spent years widening its definition of a good outcome, moving from raw watch time toward satisfaction, adding survey responses, returns, and shares to what it once counted in minutes.
Underneath the differences, these systems are all making the same move. They’re shifting from measuring whether someone was exposed to something toward measuring whether anything happened afterward.
Exposure is a cheap signal. Consequence is an expensive one. Culture is made entirely of consequence.
Content Is an Object. Culture Is a Behavior.
This is the distinction I keep coming back to, because it explains almost everything else.
Content is something a person watches. Culture is what starts happening because they watched it.
The distinction gets easier to use when you put a measure under it. The definition we work from is that participation is the voluntary time and effort invested in a brand. Watching costs time. Making your own version costs time and effort. Both are voluntary, and the voluntary part is what separates either one from reach, which nobody chooses.
A video is content. The dance thousands of people perform afterward is culture. A phrase is content. The moment people start using it on each other, it becomes culture. A brand makes a post, and that’s content. When people begin making their own versions, sending it to friends, turning it into a joke, adopting its language, or wearing it as a badge of who they are, something else is going on.
The clearest example I know is e.l.f. Cosmetics. In 2019, the brand, working with Movers+Shakers, did something almost no advertiser had tried. Instead of making a film, it commissioned an original song called “Eyes, Lips, Face” and released it into TikTok as a challenge.
The song was content. What followed was culture. More than five million people made their own videos using it. It became the fastest campaign to reach a billion views on the platform at the time, hitting that mark in six days. Then people started searching for the song on Spotify, so the team extended it into a full track. It reached number four on Spotify’s global viral chart and picked up roughly 300,000 playlist adds.
That’s the tell. The asset escaped the campaign. It stopped being an ad people watched and became a thing people used, which is an entirely different category of object.
The algorithm didn’t understand any of this philosophically. It never sat there and concluded, ah, culture. But it could see the behavioral exhaust culture leaves behind, the sending and responding and recreating and searching and returning and joining. Those behaviors reliably indicate that something mattered, so the systems distributed the thing that caused them.
Algorithms have accidentally become culture-detection machines.
Finished Work Has Nowhere to Go
This explains something brands have been frustrated by for a decade.
Most brand content is designed to be finished. We develop the message, craft the execution, polish the film, publish the post. Here it is. Please consume it. The audience has no role in the work other than receiving it. Then it gets stored on Ads of the World, and everyone moves on.
By traditional advertising standards, that can be a total success. The communication is clear, and the craft is extraordinary. Judged as a broadcast artifact, it works.
Social platforms ask a question broadcast never had to answer: What happens next?
Does someone send this to another person? Does it give them something to say? Does it become useful in a conversation they were already having? Does anyone want to add to it?
A great deal of brand content has no answer, because it arrives complete. Complete things are difficult to participate in because they contain no invitation.
The work most capable of producing culture shares a few qualities, and none of them are about production value.
It leaves an opening. There’s something for another person to contribute, complete, interpret, or extend. The e.l.f. song wasn’t a finished message. It was equipment other people could perform with.
It touches identity. People use culture to say something about themselves, about what they believe and where they belong. Spotify Wrapped is the purest commercial version of this. Spotify doesn’t make an ad. It builds a format the audience performs, once a year, on their own feeds, at their own expense, using their own data as the content. The brand supplies the template. Millions of people supply the media and the distribution.
It creates social currency. Sending something to a person communicates more than “watch this.” It might mean this is us, or you need to see this, or I can’t believe they said this, or this made me think of you. Every one of those is a reason to hit send, which is exactly the signal Instagram says it weighs most heavily.
It travels without its creator. A format, a phrase, a sound, a ritual, or a behavior can get picked up and carried by people who don’t care who made it. That’s a feature.
It connects to existing culture. Rather than demanding the world pay attention to a brand’s conversation, it earns the brand a role in conversations already happening.
None of this requires low production values. Beautifully made things create culture all the time. The mistake is believing beauty alone makes something move. Craft creates the thing. Participation creates the movement around it.
The Engagement Bait Problem
Participation doesn’t mean ending every post with “what do you think?” It doesn’t mean manufacturing controversy because comments are good for reach. It certainly doesn’t mean engagement pods, reply farms, or tricks designed to simulate activity.
Platforms are getting better at spotting all of it, and they’re increasingly willing to act. Instagram has said it removes accounts that repost other people’s content ten or more times in thirty days from recommendation surfaces entirely, and that when duplicates exist it recommends only the original. The arbitrage window on fake behavior keeps narrowing.
The objective is meaningful human consequence. Did this give people something worth doing?
Voluntary is the word carrying the weight in the definition. Bait extracts a response. Participation earns one, and the difference shows up eventually in whether anybody comes back.
That’s a much harder creative problem than making an ad. It’s also a far more interesting one.
What Changes in the Room
This changes the questions the industry needs to ask itself.
The old question was what we want people to know. That’s still important if there is genuine news or an introduction to make.
But now add another. What could this cause people to do?
Instead of designing the message and handing it to media, design the behavior around the message. People’s response to the work has become part of its distribution rather than a report filed afterward.
Instead of a content calendar full of things the brand intends to publish, build a view of the cultural moments, communities, and behaviors the brand could join or enable.
And instead of measuring only how many people something reached, study what happened because it reached them. Sends per reach is a more honest creative scorecard than impressions ever was.
In a participatory world, the audience has stopped being the destination of the media. The audience is part of the media system.
For decades, one of marketing’s central jobs was crafting a message and finding efficient ways to deliver it. That job hasn’t disappeared. But a second job has arrived alongside it. Creating the conditions under which people want to carry the idea themselves.
That demands a different toolkit. Cultural fluency. Community understanding. Speed. Generosity. A tolerance for unpredictability. And, hardest of all for most brands, a willingness to surrender a little control over what the work becomes.
The CMO is still crafting the message. They’re also building the playground.
Culture Has Always Worked This Way
Which is why this is bigger than a social media observation. It’s another instance of the larger shift from persuasion toward participation.
The traditional communications model imagined value moving in one direction. Brand to message to audience. The emerging model is messier and more alive. Brand to person to person to community to creation to reaction to adaptation, and back to the brand again.
So the most valuable thing a brand makes may no longer be the piece of work. It may be the human activity the work sets in motion.
That’s culture, and culture has always behaved this way. It doesn’t arrive fully formed from institutions. It gets repeated, altered, argued over, performed, and passed from one person to another.
Algorithms didn’t invent that behavior. They built the first distribution system large enough to see it.
Brands that understand participation now have a structural advantage rather than a stylistic one. The algorithm doesn’t want your content. It wants evidence that your content mattered to somebody, and the strongest evidence available is that they did something with it.
Which is the bet worth making. The most valuable brands will be the ones that have earned the largest groups of active participants, and the measure of a participant is what they were willing to give you that nobody asked them for.
Content gets distributed. Culture gets replicated.