A teachable system for paid creative

Turn one winning ad into ten.

Most accounts don't fail because they run out of ideas. They fail because they don't know how to multiply the idea that already works. This is the engine for doing exactly that — built so anyone, with zero background, can run it.

3levels of change
30+creative formats
15new creatives / week
The mental model

Think of it like cars.

Every change you make to a creative falls into one of three buckets. The car analogy is the fastest way to feel the difference — and once it clicks, you'll never confuse a real test for busywork again.

01

Iterations

Same car, different specs

It's still the same Lamborghini Huracán — you're just swapping the colour, the wheels, the body kit. The core idea is untouched. You're changing the surface to find a sharper version of a thing that already works.

Same idea · new surface
02

Variations

Different kind of supercar

Not the Huracán anymore — now it's a McLaren, a Ferrari, an Aventador. A different car, but the same experience is expected. Same message, brand-new creative style.

Same idea · new format
03

New Messaging

Truck vs minivan vs supercar

A wildly different experience altogether. New audience, new awareness level, new problem, new emotion. This is where you discover the next thing worth multiplying.

New idea entirely
The formula that defines each level
Awareness stage · Persona · Sophistication stage · Problem or desire · Emotion you're appealing to. Iterations and variations keep all five fixed. New messaging changes at least one.
Level 01 · Iterations

Same car. New colour.

You found a winner. Don't reinvent it — refine it. An iteration keeps the creative style and the message identical, and changes one surface-level element with a clear reason behind it. Every iteration has an intention. You are never just spamming background swaps.

What stays locked

  • Creative style
  • Awareness stage
  • Persona
  • Sophistication stage
  • Problem or desire
  • Emotion you're appealing to

What you're allowed to change

  • Location / background
  • The person (age, gender, ethnicity, look)
  • Small word tweaks in the hook/headline
  • Longer vs shorter cut (remove, never add)
  • Captions on/off · camera angle · visual hook
  • Add B-roll, a testimonial, or a stronger CTA

Worked examples · one change at a time

Different
location
Before
"Here's how we scale Facebook ads" — filmed at the office desk.
After
Exact same script, filmed driving a rented Lamborghini.
Why  Location borrows authority. Same words, ~5× the views — the only variable that moved was the backdrop.
Different
person
Before
Winning creative featuring an 18-year-old woman.
After
Same creative, same script — a 34-year-old man instead.
Why  The account skewed female. A male face opens a new slice of the same audience without changing the message.
Different
age bracket
Before
Creative built around a 20-something talent.
After
Same script and style, delivered by someone in their 50s.
Why  If the product spans ages, the talent's age is a lever. Same message, a face the older buyer sees themselves in.
Small word
tweak
Before
Hook reads "Attention Americans…"
After
Hook reads "Attention seniors…"
Why  One word, sharper specificity. CPL dropped from ~$45 to ~$35. Nothing else touched.
Shorter
cut
Before
A 30-second winning video.
After
Same video, with a slow 5-second block cut out.
Why  You only remove, never add. Add too much and it stops being an iteration and becomes a variation.
Captions
on/off
Before
Winning video running with no on-screen captions.
After
Same video, burned-in captions added throughout.
Why  Most of the feed watches on mute. Captions can lift retention on the exact same edit — one deliberate change.
Different
camera angle
Before
Talent shot straight-on at eye level.
After
Same script, shot from an unusual low or off-centre angle.
Why  A fresh angle others aren't using can stop the scroll on its own — message and style untouched.
Add a
testimonial
Before
Winning video makes a claim about the result.
After
Same video with a short customer testimonial dropped in to back the claim.
Why  Adds belief around an idea you're already conveying — strengthening the same message, not replacing it.
We iterate on winners by default — to find a sharper version of what's working. We only iterate on a loser when there's something blatantly obvious to fix.
Level 02 · Variations

Same experience. Different car.

You've found winning messaging. Now you stress-test it across brand-new creative styles. The audience, problem and emotion stay the same — only the vehicle delivering them changes. This is where one message becomes a fleet of creatives.

What stays locked

  • Awareness stage
  • Persona
  • Sophistication stage
  • Problem or desire
  • Emotion you're appealing to

What you're allowed to change

  • The entire creative style / format
  • UGC vs long-form text image vs static
  • A different hook that spawns a new script
  • Photo-led vs video-led delivery
  • Borrowed structures from other niches

Worked examples · same message, new vehicle

Static →
Us vs Them
Winning message
"Viral new laser treatment vanishes acne in 30 minutes."
New style
Same message, rebuilt as a side-by-side comparison vs traditional acne meds.
Why  Same promise, framed competitively. A new creative style reaches people the original layout didn't.
Cross-niche
hijack
Found elsewhere
"POV: you trusted the cloud pillow and the stiffness is gone" — a pillow ad.
Adapted
"POV: you tried the viral laser treatment that vanishes acne" — same structure, your product.
Why  Both sell a desired end-state. Borrow the winning structure from any niche — the one rule: never copy a direct competitor.
Photo →
video
Winning message
A static image carrying the winning text hook.
New style
Same text hook opening a video that walks through close-up results.
Why  Variations aren't limited to image-vs-image. The same message in motion reaches a different viewer entirely.
The research move: open a creative library, filter to high-traffic shops in any niche, and study their longest-running ads. Then bring those structures — not their words — back to yours.
Level 03 · New Messaging

A different vehicle entirely.

This is the only level where you change the formula itself. New awareness level, new persona, new sophistication stage, new problem, or new emotion. New messaging is how you discover the next winner — which then earns its own iterations and variations.

What a new message changes

  • A new awareness level
  • A new persona / audience
  • A new sophistication stage
  • A new problem or desire
  • A new emotion to evoke

How you test it

  • Keep the first test simple — a basic static
  • Run it against your current control
  • If it wins decisively, it becomes the new core
  • Then layer variations on top
  • Then iterate the winning variation

Worked example · two angles, head to head

Message
A
Angle
"Viral new laser treatment vanishes acne in 30 mins" — speed & novelty.
Result
Won decisively. Became the account's core message.
Why it's new messaging  Different emotion and angle than B — not a restyle of the same idea.
Message
B
Angle
"Forget Accutane — laser acne away" — positioned against an existing solution.
Result
Lost. Retired the angle, kept the learning.
Why it's new messaging  Comparing to another treatment shifts the sophistication stage — a different formula entirely.
Level 03
New messaging
Find the angle that wins
Level 02
Variations
Restyle the winner
Level 01
Iterations
Refine the best style
Then
Formats
Multiply across the library

That's the full evolution: discover the message, restyle it into variations, iterate the best one — and then explode it across the formats below.

The fourth multiplier

The format library.

Once you have a winning message and a winning style, formats are how you scale even further. The same winning idea can be re-expressed across dozens of formats — each one a fresh way to put the message in front of a different kind of viewer. Tap any card for the breakdown and a folder of real examples.

All formats
Video
Image
Carousel
Calibration

What is not a real test.

The fastest way to fool yourself is to feel busy while changing nothing meaningful. These look like tests but don't belong to any of the three levels — they don't change the message, the style, or the formula. The thread running through all of them: no intention, no signal.

Rephrasing the same hook

Saying the identical thing a different way, or just restyling the words — "3 reasons why" then "here are the reasons why," or the same line in a new font. The promise, angle and meaning are unchanged, so there's nothing new to learn.

What would count insteadA genuinely different hook that reframes the offer — e.g. shifting from a benefit hook ("clear skin in 30 mins") to a problem hook ("still hiding your acne in photos?"). New angle, new thing to measure.

Changing background colours

Recolouring the background or swapping a tint with no reason behind it is cosmetic noise. Compare to the iteration section: changing the background with intent — added authority, a new context — is a real test. Same action, the intention is what decides.

The deciding questionCan you state why this colour should perform differently? If not, it's decoration, not a test.

Cosmetic-only changes

Nudging a logo, resizing text, tidying alignment — polish that doesn't touch the message, the style, or who it speaks to. It can make a creative look better, but it isn't a variable you're testing.

What would count insteadChanging the creative style entirely (static → UGC), or the message (new angle) — something with a hypothesis attached.

Random background swaps

Generating ten background variants with no hypothesis behind any of them. Volume without intention is just noise you'll have to wade through later — and you won't be able to read what actually moved the result.

What would count insteadOne background change with a reason: "filmed in a clinic to add authority" vs "filmed at home to feel relatable." Now the swap is a question with an answer.

Copying a direct competitor

Lifting a competitor's exact creative isn't a variation — it's a liability. You inherit their mistakes, you can't out-execute your own clone, and you're legally and creatively exposed. Borrowing a structure from another niche is smart; copying a rival in your own is not.

The lineBorrow structures from other niches and make them yours. Never copy a competitor in your own category.

Renaming or re-uploading

Changing the ad name, or duplicating the identical file to "reset" it. The audience never sees either move — it's the same car with a new licence plate. No new creative, no new signal.

Also not a testSwapping the CTA button label (Shop Now vs Learn More) — that's a setting, not a creative variable.
The test for a real test
Ask: "Did I change the message, the creative style, or the formula — and can I say why?" If the answer is no on all three, you haven't run a test. You've moved a deck chair.
Best practice · how much, how often

The weekly rhythm.

Discovery is exciting; doubling-down is what scales. The 80/20 rule keeps the ratio honest — most of your energy refining what works, a steady trickle finding what's next. Here's how that turns into a simple, repeatable week.

Monday
Iteration1 × iteration
Tuesday
Iteration1 × iteration
Wednesday
Variation1 × variation
Thursday
Variation1 × variation
Friday
New msg1 × new messaging

The 80 / 20 split

80% · double down
20% · explore

Eighty percent of effort goes to iterations & variations of things you already know work. Twenty percent goes to testing genuinely new messaging. At scale, the win isn't more new angles — it's getting clearer and more specific on the one message that converts most.

Ad ideas per week5
One idea = one ad set5 ad sets
Creatives per idea× 3
Per week15 creatives
Every entry in the roadmap carries a hypothesis. Even inside the 80/20, every test is intentional — you write down which level it is and what you expect to learn. Want more volume? Bump the multipliers. But the moment quality slips, that's the signal to grow the team, not the output.
Production cycle

Roughly two weeks: week one is research + briefing, week two is production. After the first cycle it becomes a revolving door — new ads ship the moment they're ready.

One idea per ad set

Keep testing clean: one ad idea per ad set. Five ideas a week means five new ad sets, each a single line on the roadmap.

Source content honestly

Don't rip content. Everything comes from AI or hired creators. Volume is capped by the cash flow for content and team — not by ambition.