Strategy guide
How many ad concepts should a small business test each month?
For most small businesses, three to five genuinely different concepts a month is a sensible start; ten suits several products or audiences, and twenty suits multiple brands or campaigns. This is a working guide, not a Meta rule: the right number is however many distinct ideas you can run and learn from.
What counts as a concept
A concept is a distinct idea about why someone should care: a different benefit, a different objection answered, a different audience motivation or a different offer. A new colour or headline on the same idea is a variation, not a new concept. Ten variations of one idea teach you much less than five different ideas.
A practical range
| Your situation | Concepts a month | Matching plan |
|---|---|---|
| New to paid ads, one main offer | 3 to 5 | Starter (5 concepts) |
| Several products, services or audiences | 8 to 10 | Growth (10 concepts) |
| Multiple brands, locations or campaigns | 15 to 20 | Scale (20 concepts) or a custom plan |
These ranges are Moonlighters' working guidance, not a Meta requirement or an industry benchmark.
A budget-based way to work it out
The number of concepts you can usefully test is limited by how much spend each one needs before you can judge it. Two rules of thumb circulate in the industry. Both come from agencies and tool vendors, quote US dollars, and are built for larger budgets than most small NZ advertisers run, so use them as a sanity check, not a target:
- Spend-based tiers. One agency analysis of 80 direct-to-consumer accounts, as cited by the analytics vendor Segwise, suggests 4 to 8 new concepts a month under US$5,000 of monthly ad spend, and 8 to 15 between US$5,000 and US$20,000. A practitioner rule quoted in the same article is about one new concept per US$5,000 of monthly spend below US$50,000.
- Cost-to-decide. The RocketshipHQ creative testing calculator estimates that each concept needs roughly twice your target cost per acquisition in spend to reach a keep-or-kill decision, and that a share of concepts, not all, will beat your current best. It states a 15% win rate as an example, meaning about 34 concepts to find 5 winners. It gives no data source for these figures.
A worked example (illustrative numbers)
Suppose you spend NZ$1,500 a month on Meta ads and set aside 20% for testing, which is NZ$300. If your target cost per enquiry is NZ$30, then at roughly twice that per concept (NZ$60), your test budget covers about 5 concepts. If your target cost per enquiry is NZ$100, it covers only about 1 to 2 concepts, so you would test fewer and give each more time. The 20% share and the 2x multiplier are assumptions you should replace with your own.
None of these figures is a Meta rule, and the sources describe no methodology, so treat them as starting points. Your own results, tracked over two or three months, are the better guide.
Give every concept a job
Our framework assigns each concept a role so a batch covers different jobs rather than repeating one:
- Introducer: reaches people who do not know you yet.
- Scroll-stopper: a strong image or line built to earn attention.
- Problem / solution: names a problem customers recognise, then shows the fix.
- Before / after: makes the result visible at a glance.
- Objection handler: answers the doubt that stops people buying.
- Differentiator: explains why you and not the business down the road.
- Social proof: lets your genuine reviews and customers do the talking.
- Offer: gives people who are ready a clear reason to act now.
How to decide how many to run
- Start from budget. Spread across many concepts, a small budget gives each one too little exposure to tell you anything. Fewer concepts with enough spend each are easier to read.
- Test a few at a time. Launch a small set, leave them long enough to gather results, then replace the weakest.
- Change one big thing. If two concepts differ in five ways you will not know which difference mattered.
- Keep what works and refresh. Audiences get tired of the same ad, so plan a steady stream of fresh concepts rather than one big batch.
- Feed the results back. Which angle won should shape next month's roadmap.
Where finished files come in
Each concept needs a Feed version and a Story version, so five concepts is ten finished files. See the Meta ad size guide for the formats.