Creative is the biggest lever left in paid social. Targeting is mostly automated, bidding is mostly automated, but the ads people see are still your decision. The problem is that most teams test creative at random and learn very little from it.
Test concepts before details
A concept is the big idea of an ad: the problem it shows, the promise it makes and the type of person on screen. Details are things like colors, music and button text. Always find a winning concept first, then improve the details.
The weekly testing loop
- Monday: review last week’s results and pick winners and losers.
- Tuesday: write briefs for three to five new concepts.
- Wednesday and Thursday: produce the new ads, including UGC and AI variants.
- Friday: launch the new tests in a dedicated testing campaign.
What to measure
- Hook rate: the share of people who watch the first three seconds.
- Hold rate: the share who keep watching to the middle of the video.
- Click-through rate: whether the ad creates interest.
- Cost per paying user: whether the ad brings the right people.
A creative with a great click-through rate that brings the wrong users is not a winner. Judge every test on the event that makes you money.
Move winners into scale campaigns
When a test beats your current best ad on cost per paying user for a full week, move it into your main campaign. Retire ads whose results drop for two weeks in a row, before they start to drag the whole campaign down.
How many new ads do you need?
As a rough guide, larger budgets need more new ads each month, because each ad tires faster when it is shown to more people. Plan production around your spend, not around how many designers you have.
We produce and test creative every week for our clients. See how our performance creative and AI creative factory work.
