CHANNEL BUILDING

HOW WE APPROACH IT

A channel starts with a question about supply, not demand. Almost every audience is already being served something; the useful question is what they are being served badly. A topic covered only by long, slow videos when the audience clearly wants short ones. A market where every channel in a category publishes in English to viewers who would rather watch in their own language. A format that works on one platform and nobody has bothered to rebuild for another.

We look for those gaps deliberately, and we look for several at once. New channels launch in batches rather than individually, because the failure rate is high enough that a single launch tells you almost nothing. Each one gets a defined test window, a publishing commitment it has to meet, and benchmarks agreed before the first upload goes live.

What we are watching in that window is rarely subscriber count. Early subscriber numbers are noisy and easy to mistake for progress. Retention curves, returning viewers and the shape of the comments tell us far more about whether an audience is forming or whether a video simply got lucky in recommendations once.

Format before content

The instinct on a new channel is to think about topics. In practice the format matters more – length, pacing, structure, how the first ten seconds work, whether there is a face on camera. Two channels covering identical subject matter with different formats will perform nothing alike.

So the first weeks are format testing rather than content planning. We publish variations deliberately, hold the topic roughly constant, and let the retention data indicate which structure the audience actually wants. Once a format proves itself, it becomes a template the team can produce against repeatedly, which is what makes consistent output possible at all.

Consistency is the hard part

Every channel that works has one thing in common, and it is not creative brilliance. It is that something went out on schedule for months before the results justified the effort. Algorithms across every platform reward reliability, and audiences form habits around it.

This is where most channel projects quietly fail. The first eight weeks produce very little, motivation drops, publishing slips, and the channel never reaches the point where compounding begins. We plan for that period explicitly: production is batched in advance so output does not depend on anyone’s enthusiasm in a given week, and channel managers are accountable for schedule adherence separately from performance.

Built for the market, not translated into it

A channel aimed at viewers in Germany is not a translated version of one aimed at the United States. Pacing expectations differ, humour differs, what reads as credible differs, and thumbnail conventions differ enough that a straight port usually underperforms badly.

Where a market justifies it, we build separately – local voices, local editing conventions, local references. AI-assisted localisation lets us adapt production at a volume that would otherwise be impossible, but the judgement about what needs rebuilding rather than translating stays with people who know the market.

WHAT MAKES IT WORK

A willingness to close things.

Most of the channels we start do not survive their test window. That is by design, and the discipline to shut something down at week ten – while it still feels like it might turn around – is what keeps production capacity available for the ones that deserve it. Channels are not projects anyone becomes attached to. They are experiments with a defined budget and a defined question.

Production capacity, not individual talent

A channel that depends on one exceptional creator is fragile. We build production systems instead: documented formats, shared editing standards, a bank of tested hooks and structures, and enough people trained on each format that no single absence stops publishing. AI handles a growing share of the volume work in editing, localisation and variation, which raises how much any one team can sustain.

The result is that a proven format can be extended into a second or third channel far faster than the first one took to build, because the hard part – figuring out what the format should be – is already done.

Organic and paid inform each other

Because we run both, each tells us things the other cannot. Paid campaigns reveal within days which angles and hooks an audience responds to, which shortens the format testing phase on a new channel considerably. In the other direction, an organic channel with real retention data shows us what holds attention past the first few seconds, and those insights go straight back into creative for paid.

Teams that only do one of these are working with half the information.

Ownership changes the time horizon

A channel built for a client is built to show results within a contract period. A channel built for ourselves has no such deadline, and we can therefore make decisions that look poor for two quarters and correct over two years – rebuilding a format that is working adequately, investing in a market before it is obviously worth it, holding a channel through an algorithm change rather than abandoning it.

Over eight years the channels that produced the most were rarely the ones that looked most promising in month three. That is the whole argument for building assets rather than running campaigns, and it only works if you are prepared to wait.