A TikTok campaign needs 8 to 20 live ad variations in rotation, and because creative on TikTok is effectively dead within about three weeks, sustaining a 12-variation rotation means shipping four to six new cuts every week. That number is arithmetic, not ambition: divide the size of your rotation by the working lifespan of a creative in your vertical, then add a margin for the tests that fail. Most teams discover they have been running a monthly cadence against a weekly decay rate.
Why TikTok needs a faster refresh than Meta
The decay figures we plan against come from our 2026 fatigue benchmark, a synthesis of published platform and agency sources rather than our own measurement. The shape of the curve is consistent: CTR declines 15 to 20 percent in a creative's first two weeks, week three is a cliff at 45 to 70 percent below the launch baseline, and by week five a creative averages 38 percent below its peak. Most creative is effectively dead within three weeks.
TikTok compounds that in two ways. Its median video CTR is about 0.84 percent against Meta feed's 1.62 percent, so you need more impressions to buy the same volume of clicks, which pushes frequency up faster at the same budget. Its frequency ceiling sits at about 3.0, slightly more forgiving than Meta's 2.5, but not enough to offset the difference. In practice we run a weekly refresh on TikTok and a fortnightly one on Meta feed.
The exposure figure worth keeping in your head: click-through drops 45 percent after a fourth exposure to the same creative, from Meta's own internal research. Once frequency crosses 3, a large slice of your delivery is going to people who have already decided.
The weekly variation arithmetic
One formula, three inputs.
Weekly new variations = live rotation size / creative lifespan in weeks
Then divide by your hit rate if you are counting winners rather than ships, because not every variation earns a place in the rotation. Most accounts keep somewhere between a quarter and a half of what they test.
Lifespan varies more by vertical than by anything else you control. The benchmark's days-to-a-40-percent-CTR-decline figures give a usable proxy for when a cut needs replacing.
| Vertical | Days to 40 percent CTR decline | Lifespan in weeks | Weekly ships for a 12-cut rotation | Weekly ships for a 20-cut rotation |
|---|---|---|---|---|
| Food and beverage | 9 | 1.3 | 9 | 15 |
| Fashion | 12 to 14 | 1.9 | 6 to 7 | 11 |
| Beauty and DTC | 18 | 2.6 | 5 | 8 |
| Electronics | 21 | 3.0 | 4 | 7 |
| B2B SaaS | 28 | 4.0 | 3 | 5 |
Two honest caveats on that table. It assumes you replace at the decline point rather than at death, which is the right call because waiting until a creative is 45 percent down means paying for the decline. And a 20-cut rotation is not automatically better than a 12-cut one; it is better only if you can feed it, because a starved 20-slot rotation is a 6-slot rotation with fourteen zombies in it.
If you want to run your own numbers against your spend and frequency rather than these defaults, the ad variation calculator does the same arithmetic with your inputs.
Sanity-check it against frequency, not against the calendar
The formula gives you a plan. Frequency tells you whether the plan is working. If weekly frequency is drifting above 3 on TikTok, your rotation is too small for your budget regardless of what the table says, and the fix is more variations rather than more budget on the ones that are working. That is the counter-intuitive part: a rising CPA with stable hook rate and completion rate is a volume problem.
Hook-first variation, and where it stops working
Four to nine genuinely new concepts a week is not achievable for most teams, and it is not necessary. The structure that is achievable: a small number of new concepts, each fanned out into several hook variants.
A concept is a body. The evidence, the demonstration, the argument. A hook is the first one to two seconds. Swapping hooks against a fixed body is the cheapest variation there is, and it targets the metric with the most variance, so it is where the weekly number should mostly come from.
A working weekly shape for a beauty or DTC account needing five ships:
- One new concept, built from footage not yet used in this campaign.
- Four hook variants across the two concepts already performing.
- Retire the two oldest cuts, whatever their numbers look like this week.
Where hook-first stops working is worth stating plainly, because it is the trap in this approach. Our benchmark report's operational conclusion is that the number of unique creative concepts a team ships per month predicts campaign longevity better than the quality of any single ad, and brands shipping 15 to 50 variants a month see 3 to 5 times longer campaign lifespan than quarterly refreshers. Fifteen hook variants on one body is not fifteen variants as far as the audience is concerned. The footage is the same, the pacing is the same, and fatigue arrives on schedule for all of them together.
So run a ratio and hold yourself to it: at least one new concept per week, never more than about five hook variants per concept. If a month goes by with no new body footage in the rotation, the account is coasting even if the ship count looks healthy.
| Variation type | Effort per unit | Variance it moves | Share of weekly output |
|---|---|---|---|
| New concept, new footage | High | Everything | 20 to 30 percent |
| Hook swap on a proven body | Low | Hook rate, biggest single lever | 40 to 50 percent |
| Length recut, 12s and 18s | Low | Completion rate | 10 to 20 percent |
| CTA and last-three-seconds change | Very low | Click-through | 10 to 20 percent |
| Voiceover or caption style swap | Low | Modest, mostly a tiebreaker | Up to 10 percent |
What five variations a week actually costs
Two currencies: hours or credits.
In hours, a genuinely new cut from existing footage typically takes an editor 45 to 90 minutes once the footage is logged, and a hook swap on an existing timeline takes 10 to 20. Five ships a week comes out at roughly three to five hours of editor time, plus scripting and review. Manageable for one account, not for four.
In credits, our own pricing gives the other version of the sum. Genyad is our product and it is credit-based with no subscription: one standard variation costs one credit. Five a week is about 22 a month, which sits inside the Growth plan at €99 for 65 credits and leaves room for a second campaign. Starter at €29 for 15 credits covers about three weeks of a five-a-week cadence, which is the right size for testing the approach on one campaign rather than running an account on it. The free plan gives five variations plus one AI-generated video with no card, which is exactly one week of the cadence above. Credits never expire, and editing, re-exporting and uploading footage cost nothing.
The reason the credit number is lower than the hour number suggests is the part of the workflow we changed. You upload your footage once, it gets transcribed and tagged, and each variation is a fresh script, shot selection, voiceover and caption set built from that library rather than a re-cut of the same timeline. That matters here: fifteen re-cuts of one edit fatigue as a group, because the viewer recognises the shot sequence before the script. The TikTok ad maker page has the detail.
The limits, since this is a post about volume: no AI avatars or synthetic presenters, no static formats, no predicted performance score to tell you which of the five will win, and no direct publishing to TikTok. And if your footage library is shallow, volume will hurt rather than help, because five variations from twelve clips are five views of the same twelve clips.
Frequently asked questions
How many TikTok ads should be running at once?
Plan on 8 to 20 live variations per active campaign. Below eight, one fatiguing creative drags the whole ad group; above twenty, most accounts cannot produce fast enough to keep the slots fresh and the extras sit there accumulating frequency. Pick the number you can actually feed weekly.
How often should TikTok ad creative be refreshed?
Weekly for most accounts. Creative loses 15 to 20 percent of its click-through in the first two weeks and falls 45 to 70 percent below baseline in week three, so a monthly cadence means running most of your budget through declining creative. Food and drink needs replacement fastest, at around nine days, and B2B SaaS is the slowest at around 28.
Is it enough to just change the hook for a new variation?
For most of your weekly output, yes, and it is the highest-leverage change you can make cheaply. But hook swaps on the same body share the same footage and pacing, so they fatigue together. Ship at least one genuinely new concept a week alongside them.
How do I know my rotation is too small?
Watch weekly frequency. On TikTok, drifting past about 3.0 while hook rate and completion rate hold steady means the same people are seeing the same ads too often, and the answer is more variations rather than more budget or new targeting.