Most creative testing I see is just throwing loads of ads at the wall and hoping. The problem is that when you test too many different messages at once, the data becomes unreadable. You end up with a winner and no idea why it won.
how we structure it
- One copy per pain point. We group the audience's problems into three to five angles and write one piece of copy for each. Clean data, clear answers.
- Test radically different looks, not just new products. If all your product photos look the same, the algorithm treats them as repeats of the same ad, even when the product changes. Different contexts, different angles, different formats.
- Put people in their world. For top-of-funnel ads, show your customer where they actually live their life, at the gym, at home, on the school run, so it feels like their world, not a catalogue.
- Don't be precious about polish. We test polished ads alongside deliberately ugly ones, text-message-to-a-mate screenshots and notes-app screenshots. Paid media lets content be rougher because you're paying for the reach. Native often beats beautiful.
- Founders sell. A founder in the car on their phone, talking straight to camera, regularly outperforms a big-budget shoot.
borrow what already works
We pick a north-star competitor with a huge ad library and study what they've been running the longest, because long-running ads are the ones making money. Then we adapt the angle for our brand. You don't need to invent a new idea every week. You need to find the ideas that already work and do them better.
And once a creative is winning, protect it. Some of our best-performing ads have run for a full year. Refresh the creative around it rather than killing the format.
"Sell the benefit, not the feature."
Stephen Miles
Founder, Smiles Consulting
Founder, Smiles Consulting