A done-for-you service for consumer brands that advertise on Facebook and Instagram.
Others throw darts. We design the signal.
For heads of paid social who need every test to teach them something. Meta's AI reads meaning, not taste. We read every ad the same way before it runs, then design, test and optimise the signal that decides who sees it.
Who it's for.
Heads of paid social at consumer brands.
You run plenty of ads, and Meta decides who sees them. When one wins, nobody can say why, so the next round is another guess.
The job.
Make every test show how your creative steers who Meta finds, so the next round is designed, not guessed.
What good looks like:
- Every ad built to say clearly who it's for and why they'd buy.
- Every result explained: which signal reached which buyer.
- A method you can repeat, round after round.
What's in the way
More volume, more guessing.
Meta wants lots of creative, and lots of variety. So most teams generate, hope and copy.
- Generate: pile on ads and variations.
- Hope: let the platform pick a winner.
- Learn by gut: copy what won, without knowing why it won.
It looks data-driven. It isn't.
The data tells you which dart hit. It doesn't tell you why, so you can't aim the next one.
What you get.
- Every ad designed to send one clear signal: who it's for, how they behave, what they care about, why they'd buy.
- Every ad read before it runs, the way Meta's AI reads it: meaning, not taste.
- Tests that show how each signal shaped who Meta found.
- A next round designed on purpose, from what the last one taught.
The volume and variety Meta wants. Every ad a deliberate experiment.
How we make it happen.
Their cycle
- Generate
- Hope
- Learn by gut
Ours
- Design the signal
- Test the signal
- Optimise the signal
- Design the signal. Our StoryLab maps your buyers: how they behave, what they care about and why they buy. Our Story Engine builds each ad to say one of those clearly.
- Read it before it runs. Our signal analysis reads every part of the ad, the video, the words, the people in it and the page after the click, and works out who it points at and the interests Meta will match it to.
- Test the signal. The ads run, and we see how what each one said shaped who Meta showed it to, and who bought.
- Optimise the signal. We tighten what drifted, keep what reached the right buyer, and design the next round from what we learned.
Why it's different.
- Meaning, not taste. Every ad is read the same objective way, so ads are compared on what they say, not on who likes them.
- Built on how the platform works: what Meta has published about its own systems, what it needs to stay in business (people watching, advertisers finding buyers), and how AI reads content.
- Repeatable. A framework, not a lucky hit.
What teams usually do instead: more ads, more variations, and copy whatever wins.
When it fits.
- You're feeding Meta more creative than ever, and learning less from it.
- Your winners are hard to repeat, because nobody knows why they won.
- You sell to more than one kind of buyer, and want each one reached on purpose.
The proof.
Up to 60% more return on ad spend from ads with a tight signal.
- Monthly revenue more than tripled.A pet brand.
- Cost per trial down nearly 60%, to single digits.A wellness brand, at 22× return on ad spend over five months.
- Zero to a seven-figure run rate in 90 days.A new consumer brand.
- Cost per customer cut by two-thirds.A consumer brand.
Questions.
- Can anyone really know what Meta's AI sees?
- It reads ads as meaning, and meaning can be read. We read every ad the same way, then let real results show how that reading relates to who Meta finds. Every round sharpens it.
- Isn't more volume enough?
- Volume feeds the machine. Design is what makes the volume teach you something.
- Is any of it AI-generated?
- No AI faces or voices. Every face and voice is a real person from your footage. Software, including AI, helps us find the scenes, put them in order, draft the words around them and read each ad's signal.
When it's not the right fit.
- You don't advertise on Facebook or Instagram.
- You want one polished hero ad, not a testing program.
First step · Bring your old videos back to life
Start with 5 to 10 of your videos.
We'll read the signal in what you already have. Leave your work email and we'll show you how to send them.