Video, copy and creative produced in-house for every channel and market we run – built for volume without losing the thing that makes a piece work.
Every other area of our work depends on this one. A media buying test is only as good as the number of angles it can put in front of an audience, a channel cannot publish on schedule without something to publish, and a content cluster does not rank if nobody writes it. Production is the constraint that determines how fast everything else can move, which is why we keep all of it in-house rather than commissioning it. Our team produces video, static creative, copy and localised variants across every market we operate in. The work is built for volume – many variations tested rather than one hero asset defended – and each piece is made for the specific platform it will run on, because a format that performs on TikTok needs rebuilding rather than reformatting before it goes anywhere else.
Production planning starts from a number rather than an idea. How many variations does this test need to be conclusive, how many pieces does this channel need to hold its schedule for the next six weeks, how many localised versions does this market require. Those numbers determine the format and the workflow, not the other way around – an idea that cannot be produced forty times is not useful to us regardless of how good the first version looks.
That constraint is unusual, and it changes what gets made. Concepts are chosen partly for whether they hold up under repetition: a structure that can carry many different hooks, a format that does not depend on a single location or a single person being available, a template that a second editor can execute without the original creator explaining it.
Everything is briefed against what the channel or campaign is actually trying to learn. A creative test that varies five things at once produces a winner nobody can explain and nothing that transfers to the next test. We change one variable deliberately, hold the rest, and build a library of what worked and why.
The same idea becomes a different piece of work on each platform. Opening seconds function differently, aspect ratios change what can be shown, pacing expectations differ, and audiences on each platform have learned to recognise content that was clearly made for somewhere else. A vertical crop of a YouTube video reads as imported, and performs like it.
So we rebuild rather than resize. The concept carries across; the execution is made again for the platform it will live on. This costs more than exporting three aspect ratios from one timeline, and it is the difference between content that competes on a platform and content that merely appears there.
Copy translated directly from a working English asset almost never performs. Humour does not survive it, references land wrong, and the rhythm that made the original work disappears. What transfers between markets is the underlying angle – the reason an audience cared – and that has to be rebuilt in the local idiom.
Where a market justifies it we use local voices and local creators. AI-assisted localisation handles the volume work of adapting scripts, captions and variants at a scale manual production could not reach, which is what makes it possible to run properly localised creative in ten markets rather than defaulting to English everywhere. The judgement about what needs a full rebuild stays with people who know the market.
Creative volume has always been the limiting factor in performance work. Testing properly requires far more variations than a team can produce by hand, which is why most testing programmes are narrower than the people running them would like.
We use AI throughout the production pipeline – variant generation, editing assistance, localisation, asset adaptation across formats – and it has raised our output substantially without a proportional increase in headcount. What it does not do is decide what to make. Concept, angle and judgement about whether something is actually good stay with the team, because the failure mode of automated creative is producing a great deal of competent work that nobody responds to.
Being in the same room as the people spending the money.
Our editors sit alongside media buyers and channel managers, which means feedback arrives in hours and in specifics. Not “the creative underperformed” three weeks later, but which variation lost attention at four seconds and what the buyer thinks caused it. Over a year that loop teaches a production team more about what works than any amount of best-practice reading.
Every test we run adds to a record of hooks, structures and formats with performance attached. When a new market opens or a new channel launches, the team is not starting from instinct – they are starting from several years of documented evidence about which openings hold attention in that category and which formats survive repetition.
That library is why our second channel in a category is built far faster than the first, and why a new hire becomes productive in weeks rather than months. It is also the part of the operation that would be hardest for someone to replicate quickly, because it can only be accumulated by running the tests.
Outsourced production introduces a delay measured in days at every step: briefing, revision rounds, delivery, then the same again when the first version misses. For a campaign that needs to iterate within a day, that delay is fatal – the test ends before the improved creative arrives.
Keeping production in-house means a channel that needs six more pieces this week gets them, and a campaign that needs a rebuilt hook by tomorrow morning gets that too. It is more expensive to maintain the capacity than to buy it per project, and it is the reason the rest of the operation can move at the speed it does.
Content produced for a client is delivered and forgotten. Content produced for channels we own has to hold up over years, get rebuilt as formats shift, and remain something we are willing to have represent us for as long as it stays online.
That changes what gets approved. We are not producing to fill a deliverable list against a contract – we are producing the thing that will still be earning attention in eighteen months, and the standard is set accordingly.
Whether you’re an advertiser, a network, or someone looking to join the team – drop us a message and we’ll get back to you.
