I’ve owned an agency for seventeen years, and there have always been social experts working in my shop. I’ve followed what worked when social first emerged, and I’ve stayed on top of changes in media. I could explain any of it to a client.
Then this year, a small team and I launched a YouTube show of our own creation, The Spread Sheet, and I found out the difference between knowing about media and experiencing the results myself.
We made a great show intro. Crafted and on-brand, the kind of thing you put in your highlight reel. It didn’t work. YouTube doesn’t reward a perfectly crafted show intro. Not anymore. It rewards an opening that gives somebody a reason to keep watching twelve seconds later. Our intro was doing a job the medium had stopped asking for.
That was the first correction of many, and after enough of them we started joking that the algorithm was our creative director. The joke landed because it was largely true, and because none of us liked it.
The algorithm shouldn’t be your boss. But refusing to listen to it isn’t a way to stand by your creative principles. It’s just a refusal to understand the medium.
Creative people have always worked with constraints. Advertising people even more so, as constraints define everything we do. We’re doing work for clients, not ourselves. We’re working within the parameters of a brand. More tactically, a television spot has thirty seconds. A billboard has three. Every medium carries rules that shape what good work looks like inside it, and understanding those rules has always been part of the job.
Social platforms are no different, with one exception.
Their rules are alive. They watch what people do, learn from it, and adjust what gets distributed next. A billboard doesn’t change size based on how many people looked at it. A feed does.
The Platform Wrote Part of Your Brief
It helps to be precise about what the algorithm actually is, because the industry talks about it as though it has opinions. It’s merely an instrument.
The algorithm doesn’t decide what matters. The platform does.
Every platform has business objectives. Keep people watching. Get them to come back. Encourage sharing. Reduce the stuff people abandon in two seconds. Those objectives become product priorities, the priorities become KPIs, and the algorithm gets engineered to pursue them.
Platform strategy, then KPIs, then algorithm, then distribution.
That’s a chain of command. Business goals get compressed into KPIs, KPIs get compiled into code, and by the time anyone in your building has opened a document, the platform has effectively written part of your creative brief.
You don’t have to reverse-engineer the platform’s priorities. They’ve already put them in writing. YouTube’s own creator documentation states that its systems use average view duration and average percentage viewed as signals to inform ranking, and that it looks at whether viewers enjoyed a video using likes and post-watch survey results. TikTok’s documentation describes user interactions as the signals its recommender systems use to predict interest. Instagram tells creators that how many people like, comment, share, and save a post, and how quickly, is part of what puts that post in front of people who don’t already follow them.
If a platform rewards retention, retention is in the brief. If it rewards sending, sending is in the brief. If it rewards creation or conversation, those behaviors are in there, too.
You can decide those aren’t the things your brand cares about. That’s a legitimate strategic choice. What you can’t do is make that choice and then act surprised when your work doesn’t travel.
The Algorithm Is Part of the Medium
Here’s the mistake I see most often, including in work I’ve been responsible for.
We treat the algorithm as something downstream of creativity. First we make the idea. Then we optimize it for TikTok, cut it for Reels, turn it vertical, add subtitles. All of that happens after the idea is already decided, and none of it touches the idea itself. We’re changing the costume, not the character.
Making the algorithm an afterthought leads to making production adaptations.The adaptation that matters happens earlier than that, when you’re still deciding what the work is for. Go back to our intro. We didn’t solve our opening by making it shorter. We solved it by accepting the opening had a different job than the one we designed it to do.
TikTok isn’t television turned sideways, and Instagram isn’t a digital magazine. Each platform has developed its own patterns of consumption, creation, and participation, and their algorithms reinforce those patterns every hour of every day. The algorithm is a component of the medium, not a layer sitting alongside it.
Gravity doesn’t tell an architect what building to design. But an architect who ignores gravity doesn’t get to call the collapse a creative choice.
Three Briefs at Once
None of which means you’ve got to hand the algorithm the keys, because there’s a real problem with doing that. The platform’s KPI and the brand’s KPI are different animals.
TikTok may want another minute of attention while a brand wants another decade of relevance. Instagram may reward something being sent, but a company eventually needs somebody to buy something. A platform may reward outrage where a brand requires trust.
Sometimes those goals overlap beautifully, and those are the moments worth building toward. Sometimes they’re in direct conflict, and the platform will happily reward you for damaging yourself. Chasing whatever produces engagement is an efficient way to generate a lot of activity and very little brand value.
“Give the algorithm what it wants” is abdication with a dashboard attached.
The job is to work with it, which means modern creative work has to address three briefs at the same time.
The brand brief. What are we trying to accomplish for the business and the brand?
The human brief. What would make somebody voluntarily invest time and effort in this? Participation is exactly that: the voluntary time and effort people put into a brand, and work that earns none of it gives the other two briefs nothing to carry.
The platform brief. What behaviors does this specific environment actually reward?
Ignore the brand brief, and you create attention without value. Ignore the human brief, and nobody cares. Ignore the platform brief, and nobody sees what you’ve made.
The interesting work lives where all three intersect. The part most organizations get wrong is that the intersection has to shape the idea itself, not the way the idea gets adapted after it’s approved.
None of this is as new as it sounds. Years ago we made branded shows for eBay, and the person effectively steering the creative was a strategist working next to the director. Nobody called that a loss of craft. It was just a new arrangement of roles. The platform brief is that same arrangement with one more participant. One that never comes to the meeting and never explains its reasoning.
What the Algorithm Knows That We Don’t
Creative people often resist algorithms because algorithms reward things creative people find embarrassing.
Badly shot videos outperform beautifully produced films. Ridiculous ideas beat sophisticated ones. Something made in ten minutes travels further than something developed over ten weeks.
The available conclusion is that the algorithm rewards bad creative. Sometimes the work that travels is bad, and pretending otherwise isn’t honest. But the algorithm isn’t making an aesthetic judgment. It’s reporting what people did or liked or reacted to.
If that doesn’t resonate for you, there’s a second explanation that deserves more attention. The algorithm may be exposing a gap between what professionals believe people should value and what people actually value. That gap has always existed. We just never had continuous instrumentation pointed at it before.
Duolingo is useful here, because it looks like an accident and isn’t. A small in-house social team put the brand’s green owl mascot on TikTok and let it behave in ways no brand-safety document would have pre-approved. The account grew into millions of followers with effectively no paid media behind it, and it kept working because the team read what people responded to instead of deciding in advance what the brand was allowed to be there.
That’s the algorithm reporting a fact, and a team being willing to hear it.
It doesn’t mean the audience is always right, or that popularity equals quality. Plenty of enormous numbers have done nothing for the businesses that paid for them.
What it does mean is that something happened. People responded. Creative people should be curious about why. That’s what good partners do. They show you something you couldn’t see on your own.
Learning Beats Optimizing
Part of the problem may be the language we’ve chosen.
“Optimize for the algorithm” makes the whole thing sound mechanical. It implies creativity happens first and then somebody retrofits it to satisfy a machine. No talented person wants that job, and framed that way, they’re right not to.
The better approach is to frame it as a question: what is the algorithm teaching us?
What made people stay? What did they pass along, and to whom? What did people do with our idea that we never anticipated? Which part of the work turned out to be useful to somebody, and which part did everyone ignore?
The answers are information rather than instructions. Instructions replace judgment. Information sharpens it. And sharper judgment on the next creative decision is the only durable advantage available in an environment that keeps changing its own rules.
Traditional advertising was built around finished objects. Develop the campaign, approve it, produce it, run it, measure afterward. The artifact was the deliverable, and the learning arrived too late to influence the artifact itself.
Algorithmic environments don’t work that way. You release something, people respond, the system responds to them, and distribution shifts. The work becomes part of a living system rather than a completed object.
This is where the industry tends to overcorrect. One camp turns creativity into endless A/B testing and surrenders every decision to performance data, which produces work that’s optimized into invisibility. The other insists taste alone should be sufficient, which produces work nobody sees.
The harder position sits in between. Having the conviction to put something distinctive out there, and the discipline to evolve it without sanding away whatever made it distinctive in the first place.
That’s creative intelligence at work.
The Audience Is in the Room
For decades, advertising described itself as consumer-centric while keeping consumers outside the creative process. We researched them, segmented them, tested ideas on them, then delivered finished communications to them. They were the audience for the work and never a participant in it. Research entered the room. The people it described never did.
Algorithmic media ended that arrangement. People’s behavior now affects distribution. Distribution affects what brands and creators make next. People respond to that, and the loop continues without asking anyone’s permission.
Creative, then response, then distribution, then creative again.
The audience hasn’t become the creative director. Neither has the algorithm. But both have entered the creative department, and that changes the role of the people who’ve traditionally controlled the work.
Our joke was wrong in the way useful jokes usually are. The algorithm was never running our room. It was the one collaborator that told us the truth every day and never once told us what to do about it.
The best creative leaders will operate between conviction and evidence, understanding what the brand needs, what people care about, and the behavioral system their ideas have to travel through.
The algorithm isn’t your creative director. It’s evidence that the audience is already in the room. Great creative leaders will know when to listen, when to disagree, and what to make next.