A typical active campaign needs 8 to 20 live ad variations, and that range is an output rather than a rule you adopt. Weekly impressions divided by reachable audience gives frequency, frequency across a creative's useful life gives total exposures per person, and total exposures divided by the exposures one creative survives gives the number of variations that have to be in rotation. Move the budget or move the audience and the number moves with it, which is why two accounts on the same platform can correctly land on 7 and on 20.
Why the count is a frequency problem, not a production question
Most teams ask how many variations they can make. The binding constraint is how many the same audience can absorb before the creative stops working. Our 2026 fatigue benchmark, which is a synthesis of published platform and agency figures rather than our own measurement, puts the CTR drop at 45 percent after a fourth exposure to the same creative. It also puts weekly frequency of 2.5 on Meta prospecting as the point where decline begins, with TikTok tolerating about 3.0.
Those two figures are the whole derivation. If a person can see one creative three times before its click-through collapses, then the exposures your budget buys against that audience, divided by three, is the floor on how many creatives you need live. Everything after that is adjustment for how platforms actually deliver.
How to derive the number in four steps
Step 1. Weekly impressions = (weekly budget / CPM) × 1,000.
Step 2. Weekly frequency = weekly impressions / reachable audience.
Step 3. Exposures per person over the creative's useful life = weekly frequency × weeks of useful life. Three weeks is the default, because the benchmark finds most creative is effectively dead within three weeks.
Step 4. Rotation floor = exposures per person / 3, rounded up. Three is the exposure cap the 45 percent figure implies.
Then two adjustments that separate the arithmetic from the ad account. Delivery is not even: in most accounts the platform pushes the majority of spend into two or three ads, so your nominal rotation is roughly a third of your ad list. Multiply the rotation floor by three. And you need slots for creative that is being read rather than rotated, which is typically three to six.
Live variations = (3 × rotation floor) + test slots
Worked, for a campaign spending €12,000 a week at a €15 CPM against 200,000 reachable people:
- 12,000 / 15 × 1,000 = 800,000 weekly impressions
- 800,000 / 200,000 = 4.0 weekly frequency
- 4.0 × 3 weeks = 12 exposures per person per creative cycle
- 12 / 3 = 4, so the rotation floor is 4
- 3 × 4 = 12 rotation slots, plus 4 test slots = 16 live variations
Sixteen, not because sixteen is a best practice, but because that account's reach maths asks for it. Substitute your own budget, CPM and audience in the ad variation calculator and the same four steps run against your numbers.
What the derivation gives at four different budget levels
| Account shape | Weekly budget | CPM | Weekly impressions | Reachable audience | Weekly frequency | Rotation floor | Live variations |
|---|---|---|---|---|---|---|---|
| Broad prospecting, small budget | €1,500 | €12 | 125,000 | 500,000 | 0.25 | 1 | 7 |
| Mid-market DTC | €7,500 | €14 | 535,700 | 400,000 | 1.34 | 2 | 10 |
| Tight audience, heavy spend | €12,000 | €15 | 800,000 | 200,000 | 4.00 | 4 | 16 |
| Same audience, budget doubled | €24,000 | €15 | 1,600,000 | 200,000 | 8.00 | 8 | 28 |
The middle two rows are why 8 to 20 is the range people keep quoting. It is not a convention. It is where the arithmetic lands for accounts spending five to low six figures a month against audiences of a few hundred thousand, which describes most of the performance accounts in this market. Quote the range if you like, but the rows either side of it are where the interesting decisions are.
What changes at high budget or on a small audience
The bottom row deserves a hard look. Doubling spend against the same 200,000 people asks for 28 live variations, which is past what most campaigns can rotate and well past what most teams can read. Weekly frequency of 8.0 is more than three times the 2.5 ceiling the benchmark identifies for Meta prospecting, so no quantity of creative rescues that campaign. The fix is audience: broaden the targeting, add a placement, add a market. Creative volume buys time inside a frequency ceiling, it does not raise the ceiling.
The top row is the mistake that gets made more often. That campaign spends €1,500 a week against half a million people, weekly frequency is 0.25, and nobody is anywhere near a fourth exposure. Six or seven variations is the correct answer. Run 20 and the budget splits so thinly that no single variation collects a readable impression count, so you have paid for 20 pieces of creative and learned nothing about any of them.
A small audience with a small budget is the third case. Retargeting pools of 20,000 to 50,000 people hit frequency ceilings fast on very little spend, and they are the one place where a five-variation rotation refreshed weekly beats a 20-variation rotation refreshed monthly.
How often you have to replace them
Rotation size and refresh cadence are one calculation seen from both ends. Hold N slots live, assume a useful life of L days, and you replace N × 7 / L variations a week. The benchmark's days-to-a-40-percent-CTR-decline figures by vertical make that concrete for a 16-slot rotation.
| Vertical | Days to a 40 percent CTR decline | Replacements per week | Replacements per month |
|---|---|---|---|
| Food and beverage | 9 | 12 | 53 |
| Fashion | 12 to 14 | 9 | 37 |
| Beauty and DTC | 18 | 6 | 27 |
| Electronics | 21 | 5 | 23 |
| B2B SaaS | 28 | 4 | 17 |
Those monthly figures run from 17 to 53, which brackets the benchmark's finding that brands shipping 15 to 50 creative variants a month see 3 to 5 times longer campaign lifespan than quarterly refreshers. The two numbers agree because they are the same fact approached differently: a rotation the size your reach demands, replaced at the rate your vertical decays, produces roughly that monthly volume. A food and beverage brand and a B2B SaaS company can both hold 16 slots and have completely different production problems.
What this asks of production
The derivation only helps if you can hit the replacement rate. A beauty brand needing 27 new variations a month cannot get there with one editor and a shared timeline, and this is the point the benchmark makes bluntly: throughput, not talent, is the bottleneck, and the count of unique concepts shipped per month predicts campaign longevity better than the quality of any single ad.
Genyad is our product, so treat this as disclosure. It turns video you already own into publishable variations: you upload footage once, it transcribes and tags every clip, and each variation is a fresh script, shot selection, voiceover, caption set and export drawn from that library rather than a re-cut of one timeline. One standard variation costs 1 credit, and editing, re-exporting and uploading footage cost nothing. Growth is €99 for 65 credits, which covers a 27-a-month replacement rate for two months. Batch creation of video ad variations is the workflow that maps onto the cadence table above.
What it will not do: there are no predicted performance scores, so it cannot tell you which of your 16 will win, and there is no direct publishing to Meta or TikTok, so you export and upload. There are no AI avatars and no static banner formats. If your rotation problem is that you have no footage at all, a synthetic presenter tool like Arcads, listed at €100 per month for 10 videos in August 2026, solves a different problem better than we do. Prices move, so check before you plan around one.
One more thing worth being strict about. Four aspect ratios of the same ad are one variation, not four. The ad variation glossary entry draws that line, and counting crops as variations is the most common way a team convinces itself it has 20 in rotation when it has five.
Frequently asked questions
How many ad variations should I run per campaign?
Between 8 and 20 for a typical active campaign, but derive it rather than adopt it: weekly impressions divided by reachable audience gives frequency, multiply by three weeks of useful life, divide by three exposures per creative, multiply by three for uneven delivery, and add three to six test slots. A campaign at €12,000 a week and a €15 CPM against 200,000 people lands on 16. A campaign at €1,500 a week against 500,000 people correctly lands on 7.
Does a small audience need more variations or fewer?
More, and sooner. A small audience raises weekly frequency for the same spend, which pushes exposures per person up and the rotation floor with it. Past a point the derivation asks for more variations than a campaign can sensibly rotate, and that is the signal to widen the audience instead of producing more creative.
How many new variations do I need each week?
Divide your live slot count by the useful life of a creative in your vertical. On a 16-slot rotation that is about 12 a week in food and beverage, where our 2026 fatigue benchmark puts a 40 percent CTR decline at 9 days, and about 4 a week in B2B SaaS at 28 days. Most accounts in beauty and DTC land near 6 a week.
Do different aspect ratios count as separate variations?
No. Exporting the same ad at 9:16, 4:5, 1:1 and 16:9 is placement coverage, not creative variety, and a viewer who sees the same hook in a different crop has still seen the same ad. Count distinct hooks, arguments and shot selections instead.
Is 8 to 20 variations a benchmark or a guess?
It is the operating range in our 2026 fatigue benchmark, which is a synthesis of published platform and agency figures rather than proprietary measurement. It holds for accounts whose budget and audience produce weekly frequency somewhere between about 1 and 4. Outside that band, the derivation is more reliable than the range.