How we work
The platforms want more ads, more different, all the time. Here's how we build a content engine that can keep up.
Three forces push the same way: brands want testing that's fast and cheap, the whole industry says creative volume is the unlock, and Meta rewards creative diversity. Volume, variation, diversity: three words for one pressure. The old way, a new ad means a new shoot, and the bill grows with every test.
What most teams do with that pressure
Treat content as a finished product, and chase volume.
Ideate. Brief. Recruit creators. Generate. Test. Throw away what didn't work. Start again.
- Every video is a one-off. Nothing carries to the next round.
- Ideas come from intuition and whatever other brands are running.
- When something wins, nobody can say why, so it can't be repeated on purpose.
We take a performance-storytelling lens instead.
The difference starts with how we think a story should be made, before anyone picks up a camera. Six steps, each shown on a real brand.
Step 1 · StoryLab
We start with the buyer, not the brief.
People buy for a reason. Something gets a job done well in their life, or in the life of someone they love. So every story starts there.
From the product's features and benefits, and from what customers say in reviews, we work out who buys, the job each one needs done, and the motivation underneath. Then what each buyer would have to believe before they act, and the proof that would make them believe it. Every belief is a thesis to test.
- The buyer
- Skeptical Rewards Tester. A cautious PC gamer who will only try a rewards loop once it feels verifiable, rule-clear and automatically tracked.
- The job
- When I'm about to buy a PC game key and see "earn rewards for the games you play", I want to confirm the loop is real and my achievements will be tracked automatically, so I can make a low-risk first purchase.
- What moves them
- Clear rules and automatic verification that prove the rewards are real, before they connect their account.
- The belief
- "I trust the loop more when it reads my achievements and verifies progress by itself: no screenshots, no forms, no overlay."
- The proof it needs
- An achievement pops in the game; the quest updates on its own, on camera.
A real buyer from a gaming rewards brand's StoryLab, planned from their website and one document. 15 buyers, 71 beliefs in five kinds: what they want, the pain, why it works, why this over the rest, the worry we answer.
Step 2 · The bets
We map every story, then bet on the strongest.
Before anyone says "let's do a get-ready-with-me video", we know every story this brand could tell: every buyer, every belief, every scene that could be filmed.
Then we narrow. Not by taste, and not because another brand is running it, but by how people actually decide: which beliefs are most powerful, which proofs most likely to move them, where the buyer is in their journey. The strongest bets set the shoot list and the order of the scenes.
Step 3 · Designed first
The stories, the shot list and the landing pages exist before we film.
For every bet we take, the story is written, the scenes are listed, the landing page is laid out, and we've read the audience each one will build.
- The story card: one buyer, one belief, the eight moments of the story in order.
- The shot list: each scene with who says it, what the line has to do, how long, and how to shoot it.
- The landing page: the same story continued after the click, with the gaps marked for the footage to fill.
- The targeting read: who this ad will point at, read from the plan, then confirmed on the first cut.
- One scene from a real shot list
- The problem · 4 to 8 seconds · an everyday customer. Selfie-style at a gaming desk. In the background, a monitor shows an achievements page at 100%. The creator points at the list, then back to camera with a "that's it?" look.
- What the line has to do
- Name the enemy: achievements that earn you nothing.
Nothing is left to chance on set. The creator knows the scene, the beat and the job of every line.
Step 4 · The shoot
Real people film scenes, not ads.
We recruit the creators and brief them scene by scene. Long form: a creator telling a story in scenes. Short form: monologues and read-outs that fill content gaps fast. And B-roll we ask for on purpose, because we already know which scene it overlays and which landing-page section it fills.
The struggle 3s
The worry 3s
The fix 3s
The demo 6s
The proof 4s
The ask 4s
Six scenes from five real creators, each filmed as a scene with a job. Real creator footage from one pet brand · illustration.
Tap a scene to hear it.
- Real people, never AI faces. Buyers can tell when someone means it, and so can the platform.
- Every creator's usage permission is tracked, scene by scene.
Step 5 · The packet
One shoot flies three ways.
When other agencies finish a video, it goes into the ad account. When footage comes back to us, it fills the library and flies three ways at once.
The story as filmed, launched into your own ad account, aimed at its buyer.
The scenes join everything you already have, and new first cuts build themselves.
The same scenes and B-roll fill the page the ad points to.
So you're already testing more stories than you paid to film. The original. The remixes. And the page that carries the story after the click.
Step 6 · The read
We learn through the same lens, and design the audience.
Others look at the scoreboard and copy the winner. We tie every result back to the buyer, the belief, the proof device, the creator and the shape of the story that produced it.
And before any ad runs, we read its audience: what Meta's AI will see in it and who it will show it to. The platform reads meaning, not taste: the video, the words, the people, the page after the click, all turned into who this is for and why they'd buy. We uncover that audience before the ad runs, keep the read, and compare it across every ad. Design the audience, test it, optimise it.
Return on ad spend, shown as multiples. Up to 60% more from ads with a tight audience.
See audience designStep 7 · The curve
Then it compounds.
Every round adds scenes to the library and lessons to the next bets. Early on, you get more ads from every shoot. As the library fills, each new scene unlocks far more than one story.
- One proof device, filmed once, already sits in 25 scenes of that library. Film two more versions of it and dozens more stories open up.
- The combinations run into the hundreds of thousands, so the engine does the digging: it finds the stories worth telling, and you can ask it for a story or a scene in plain words.
- The library, the labels and everything learned are yours. In the old way, that knowledge left with the agency.
Where it goes
Your engine. We run it, or you do.
Many brands want us to run the whole thing. Others want the engine and the advice. Either way, the end state is the same: a content engine rich enough that you can ask it for the next story yourself.
Send 5 to 10 videos. We cut them into scenes, read them, and show you what they can become.
A conversation about your product and your buyers. Then StoryLab, the bets, the shoot.
We run your Facebook and Instagram ads from the engine and build each round from what we learn.
Pick whichever fits where your brand is today. Each one feeds the next.
Talk to us
Start with the videos you already have.
Leave your work email and we'll show you how to send us 5 to 10 of them, or set up a conversation about building your engine from scratch.