Why creative testing matters so much now
On Meta, the algorithm handles most of the audience selection. Instead of setting interests and age groups by hand, you give the system ads, and it finds the people they work for. That makes the creative your most important lever: every ad speaks to a different part of the market. If you only have a few ads, you only reach a few parts of the market.
The rule of thumb for how many
For an ad to be judged fairly, it needs enough budget to produce a signal. A common rule of thumb is to give each ad about one to two times your target cost per purchase before you decide.
Formula: monthly testing budget ÷ (target cost per purchase × 1.5) = number of ads you can meaningfully test.
| Monthly testing budget | Target cost per purchase | Ads you can meaningfully test |
|---|---|---|
| $1,500 | $30 | about 33 |
| $3,000 | $40 | about 50 |
| $6,000 | $40 | about 100 |
Your testing budget isn't your entire ad budget. Many brands use a portion, roughly a fifth to a third, for new ads and the rest for proven winners.
According to Meta, an ad set needs about 50 optimization events within seven days to exit the learning phase. If you spread too many ads across too little budget, none of them learns properly.
Research, copy, statics and AI UGC videos as a subscription, reviewed by performance experts. From $990 a month, cancel anytime.
What to test, in this order
- Angles and buyer types: Who is the ad for, and which problem or desire is at the center? This is where the biggest differences are.
- Formats: UGC video, static, comparison, testimonial, founder story. A new format often reaches new people.
- Hooks: The same video body with three different openings. More in video ad hook ideas.
- Offers: bundle, trial pack, guarantee.
- Details: colors, headline variations, button text, but only once you have a clear winner.
How to read the results
| Metric | What it tells you |
|---|---|
| Hook rate (video) | How many people watch the first few seconds. Shows whether the opening stops the scroll. |
| Click-through rate (CTR) | How many people click. Shows whether the message sparks curiosity. |
| Cost per purchase (CPA) | What a customer costs. The metric that counts in the end. |
| Return on ad spend (ROAS) | Revenue per ad dollar. Important when order values vary a lot. |
An ad with a high hook rate but a high cost per purchase often has a strong opening and a weak rest, or the landing page doesn't deliver what the ad promises. An ad with a low hook rate needs a new opening before you throw out the idea.
A testing rhythm that works
- Upload new ads at the start of the month, spread across several angles and formats.
- Turn off the clear losers after about a week.
- Move the winners into the campaign with your proven ads.
- Have new variations built from the winners: new hooks, new people, same message.
- Write down what you learned so the next month starts stronger.
Which plan fits which budget
The AVVEN plans follow this logic: Starter delivers 20 ads a month for brands that are just starting to test regularly, Growth delivers 50 ads with iterations based on your numbers, and Scale delivers 120 ads for brands with large budgets. Run your own numbers through the formula above and you'll quickly see which plan fits.
Frequently asked questions
How many new creatives do I need per month?
It depends on your budget. A simple rule of thumb: divide your monthly testing budget by one to two times your target cost per purchase. The result is roughly the number of ads you can meaningfully evaluate.
How long should I test an ad before turning it off?
Until it has spent about one to two times your target cost per purchase, or until it's clear that it doesn't stop the scroll. One or two days is rarely enough to make a call.
Should I test many small differences or a few big ones?
Big differences first: different angles, buyer types and formats. Details like colors or button text are only worth testing once you have a clear winner.