How Often Should You Refresh Ad Creative? A Cadence Guide
Refresh too slowly and costs creep up. Refresh too fast and you never learn what works. Here is how to set a creative refresh cadence based on spend, signals and channel.
There is no single correct answer to "how often should we refresh creative?" A brand spending $300 a day on one audience will wear out ads much faster than a brand spending $50 a day. And a ChatGPT chat card that only appears when someone asks a specific question behaves differently from a Meta video served to the same people again and again.
What you can do is set a cadence based on your spend and audience size, then let fatigue signals tell you when to speed up or slow down. This guide gives you both.
Why creative wears out
Ads fatigue for two main reasons:
- Frequency. The same people see the ad repeatedly. After a few exposures, those who were going to act have acted and the rest have learned to scroll past.
- Audience exhaustion. The platform has already shown the ad to the most responsive people in your audience. Each additional impression goes to someone less likely to care.
Both show up as the same pattern: click-through rate falls, cost per result rises, and conversion rate may slip as the ad reaches less interested people.
Not every ad fatigues on the same schedule. Some winners run for months. In Meta's Ad Library, you can often find ads from established brands that have been active for 60 days or more. Advertisers rarely keep paying for an ad that has stopped working, so long-running ads are a useful signal of durable messages. Our article on why long-running ads are your best creative research explains how to use that signal.
Signals that an ad needs replacing
Watch these metrics weekly for your top ads by spend:
| Signal | What to look for | Likely cause |
|---|---|---|
| CTR decline | Sustained drop from the ad's own early baseline | Audience has seen it too often |
| Frequency rising | Average frequency climbing week over week (Meta) | Small audience or high budget |
| CPA rising | Cost per purchase rising while CPM is stable | Ad is losing persuasive power |
| CPM rising with CTR falling | Both moving the wrong way | Platform sees lower engagement |
| Conversion rate falling | Click quality dropping | Reaching less interested people |
| Comments repeating | "Seen this ad a hundred times" | Clear fatigue |
Compare each ad to its own baseline rather than to other ads. A video that started at 1.8% CTR and now sits at 1.1% is fatiguing, even if 1.1% is still higher than another ad's best.
A practical rule: if an ad's cost per purchase has risen more than 25% to 30% from its first two weeks and stays there for a week, start replacing it. Use the free CTR calculator and CPM calculator to track these consistently.
A cadence by spend level
Your cadence should scale with how fast you burn through audiences. These are starting points, not rules. Adjust based on the signals above.
| Daily spend per channel | New creatives per week | Review cadence |
|---|---|---|
| Under $100 | 1 to 2 | Every two weeks |
| $100 to $500 | 3 to 5 | Weekly |
| $500 to $2,000 | 5 to 10 | Weekly, with midweek check |
| Over $2,000 | 10 or more | Several times a week |
Smaller budgets need fewer new ads because each ad takes longer to reach saturation, and because launching too many ads at once splits limited budget so thinly that none gets enough data.
Refresh vs replace
Not every refresh requires a new concept. There are three levels:
- Iteration. Keep the winning message, change the hook, opening line, image or format. Cheapest and often effective.
- New angle. Keep the product and offer, change the reason to buy. For example, from "saves time" to "makes a better gift."
- New concept. New message, new visual approach, sometimes a new offer.
A healthy mix for most brands is roughly half iterations on proven winners, a third new angles, and the rest new concepts. Iterations keep performance steady; new angles and concepts find the next winner. The free ad variation generator is a quick way to produce iterations, and our list of 35 ad angle ideas helps when you need new angles.
Refresh cadence for ChatGPT ads
ChatGPT ads fatigue differently from social feed ads. A chat card appears below an answer when the conversation matches your context hints, which describe the situations your ad belongs in. People do not see it on a loop while scrolling; they see it when they ask a relevant question.
That changes the refresh logic in a few ways:
- Frequency is usually lower per person. Someone has to have a related conversation to see your ad, so repeated exposure builds more slowly than in a social feed.
- Context hints matter as much as copy. If performance drops, the issue may be which conversations the ad is matched to, not the creative itself. Refine hints before replacing copy.
- New buyer questions are new inventory. Adding ads that answer different questions people ask ChatGPT expands reach more than rewriting the same answer.
A sensible approach for a new ChatGPT campaign is to launch with several ads answering different buyer questions, review weekly, and add or replace a few each refresh cycle. Each chat card has a headline of 3 to 50 characters and a body of up to about 100 characters, so iterations are fast to write. For example, two variations for a travel pillow brand:
- Headline: Neck pillow that packs flat
- Body: Memory foam support that compresses to the size of a book. Clips to any carry-on handle.
- Headline: Sleep upright on long flights
- Body: Supportive memory foam neck pillow that packs flat into your carry-on. 30-day returns.
Both are original, factual, and make claims the landing page can support. Our guide on writing ChatGPT ad headlines covers more patterns.
Building a creative pipeline
The hardest part of a refresh cadence is not deciding when to refresh. It is having new creative ready when you need it. A simple weekly pipeline:
- Monday: review. Pull performance for every active ad. Flag fatiguing ads using the signals above.
- Tuesday: research. Check Meta's Ad Library for long-running ads in your niche, read new reviews, and note recurring customer questions.
- Wednesday: brief. Write briefs for next week's ads: which winners to iterate, which new angles to test. The free creative brief generator speeds this up.
- Thursday and Friday: produce. Write copy, create or edit visuals.
- Following Monday: launch the new batch and repeat.
Keep a backlog of at least one cycle's worth of ready-to-launch ads. When a winner suddenly fatigues, you can replace it the same day instead of waiting a week.
Mistakes that break a refresh cadence
- Turning off winners too early. Some ads dip for a few days and recover. Wait for a sustained trend.
- Changing too much at once. If you replace every ad in a week, you lose the baseline you need to judge new ones.
- Iterating only. Endless variations on one message eventually hit a ceiling. Mix in new angles.
- Ignoring the landing page. If every new ad underperforms, the page may be the issue. See our guide on CRO for paid traffic.
- Refreshing on a calendar alone. A schedule keeps you disciplined, but signals should drive the actual replacements.
How SecondWin handles refresh cadence
SecondWin's plans include a built-in refresh cycle for ChatGPT ads: a monthly refresh on Starter ($99/mo, 10 creatives a month), biweekly on Growth ($299/mo, 50 creatives), and weekly on Scale ($799/mo, 100 creatives). Each refresh draws on the latest long-running messages in your niche and the buyer questions your customers ask ChatGPT, and every ad is checked against OpenAI's ad policies before it goes live. Full details are on the pricing page.
Want to see what the first batch would look like? Get a free analysis of your website with your buyer questions and starting ad concepts.
FAQ
How do I know if my ad has creative fatigue?
Compare the ad to its own early performance. Warning signs include a sustained drop in click-through rate, rising frequency, and cost per purchase climbing while CPM stays flat. One bad day is noise; a week-long trend is a signal. If cost per purchase has risen by roughly a quarter or more from the ad's first two weeks and stays there, plan a replacement.
How long does a Facebook ad last before it fatigues?
It depends on budget, audience size and the ad itself. With small budgets and broad audiences, a good ad can run for many weeks or months. With large budgets aimed at small audiences, ads can fatigue within one or two weeks. Long-running ads visible in Meta's Ad Library show that strong messages can last much longer than most people expect.
Do ChatGPT ads need refreshing as often as Meta ads?
Usually not as frequently per ad, because chat cards appear only when conversations match your context hints rather than repeatedly in a feed. However, you still benefit from regular additions, since new ads answering different buyer questions expand reach. When performance dips, review context hints and landing pages before rewriting the creative itself.
Should I pause old ads when launching new ones?
Not automatically. Keep winners running while new ads gather data, and pause old ads only when signals show fatigue or when new ads clearly outperform them. Pausing everything at once removes your baseline and can disrupt the platform's learning. Rotating gradually, replacing a few ads per cycle, keeps performance steadier.