Ad fatigue: how to tell it from the problems that look like it, and what to refresh first

When an ad that used to sell starts costing more per purchase while its frequency climbs and its click-through rate slides, you're most likely looking at ad fatigue: the same audience has seen the same creative too often. Meta's documentation describes it the same way. Confirm it before you act, because the fix is a new reason to stop scrolling, not a bigger budget or a relaunched copy.

One illustrative ad over its first three weeks shows the shape. Each line starts at 100 on launch day, so three metrics with different units can share one chart:

How one ad wears out over three weeks

Illustrative ad. Each line starts at 100 on day 1; 200 means double the day-1 value.

• Frequency since launch
• Link CTR
• Cost per purchase

075150225300159131721CTR 10% down30055182

Days since launch

Show the numbers

Days since launch	Frequency since launch	Link CTR	Cost per purchase
1	100	100	100
3	125	101	99
5	150	99	101
7	175	96	104
9	200	90	111
11	217	83	120
13	233	76	132
15	250	69	145
17	267	63	159
19	283	58	172
21	300	55	182

Nothing went wrong on day 9 that anyone could point to. The budget kept buying impressions, fewer of them reached people who hadn't seen the ad yet, and a growing share landed on people who had seen it before. Fewer of those viewers clicked. With the CPM and the page's conversion rate unchanged, every lost click went straight onto the cost of a purchase.

Facebook ad frequency: what the number means

Meta defines frequency as the average number of times each person saw your ad, and it marks the number as an estimate (Ad Insights reference). In practice it's impressions divided by reach. When an ad set spends the same every day and its audience doesn't grow, impressions rise faster than reach, so frequency keeps climbing.

A back-of-envelope check shows how fast that happens. At $500 a day and a $12 CPM, an ad set buys about 41,700 impressions a day, or 875,000 in three weeks. If only 250,000 people are realistically in reach, that's an average of 3.5 impressions each. Spread the same budget over an audience four times larger and the average can stay near 1. Small audiences with big budgets wear out first.

An average also hides the spread. A frequency of 3 can mean 80% of the audience saw the ad twice while the other 20% saw it seven times, so a modest-looking average can include people who have seen the ad far too often.

Meta publishes no frequency at which fatigue begins, and a number borrowed from someone else's account says little about yours. The same frequency can be harmless on one ad and a problem on the next. What matters is whether click-through rate and cost per purchase move with it, which is why Meta's suggested diagnosis is to read frequency and CTR day by day, together with cost per action.

So set your trigger against the ad's own history, for example a three-day CTR 20% below the ad's first-week average while frequency keeps rising. Pick the number before you look at the data, write it down, and read it as a prompt to investigate, not a verdict.

Is it ad fatigue or something else?

Rising costs have several causes that look alike on a dashboard: a worn-out creative, an audience that has seen everything you run, a pricier auction, a change in your funnel, or tracking that stopped reporting sales. Each needs a different fix, and new creative only cures the first. Answer a few questions about the drop, in order, to tell them apart:

Fatigue, or something else?

Start from how widely the drop spread

1Did results fall on one or two ads, or on every ad at once?

One or two adsEvery ad at once

2On that ad, is frequency climbing while click-through rate falls?

YesNo, CTR is steady

3Does a new ad in the same ad set start strong?

Yes, it starts strongNo, it fades fast too

If you haven't launched one yet, do: its first few days answer this question.

4Did the page, offer, price, checkout or tracking change around that date?

Yes, or not sureNo

5Are CPMs up on every ad, including ads launched this week?

YesNo

Best fit

Creative fatigue: refresh the ad

One ad's frequency is climbing while its click-through rate falls, and a new ad in the same ad set starts strong. The audience is fine; this creative is worn. Write new openings or a new angle, and lower the tired ad's budget instead of raising it.

Best fit

A saturated audience: widen it

Frequency is climbing and click-through rate falling, and new ads in the same ad set fade within days too. The audience has seen enough from you, whatever you show it. Widen the audience before you pay for more creative; Meta's fatigue alert lists expanding the audience among its fixes.

Best fit

The click still works: look after it

Results fell on one or two ads, but click-through rate is steady. People still click; fewer of them buy. Check what happens after the click: revenue per visitor from that ad, the page on a phone, and the first minute of the VSL.

Read more: Video Retention Graph: How to Read a VSL Drop-Off Curve

Best fit

Check the funnel first

Every ad fell at once, and the page, offer, price, checkout or tracking changed around the same date, or you can't rule it out. Fatigue doesn't hit every ad on one day. Compare the days before and after the change, and place a real test order.

Best fit

The auction got pricier

Every ad fell at once, nothing changed on your side, and CPMs rose on every ad, new ones included. That's the market, not tired creative, so new ads won't fix it. Recheck your break-even and decide how much volume is worth paying for.

Best fit

Check the data, then the audience

Every ad fell at once, nothing changed on your side, and CPMs are flat. Compare the platform's purchases with your own orders for the same days: a broken pixel or Conversions API feed makes every ad look worse at once. If the numbers agree, your ads may all be drawing on one saturated audience.

The first question does most of the work. Fatigue builds ad by ad, as each creative's audience sees it again, so it rarely hits every ad on the same day. When everything drops at once, look outside the creative first. A price change, a new VSL cut, a checkout update or a broken purchase event makes every ad look worse overnight. Before you touch a single ad, compare the platform's purchase count with your own orders for the same days.

The second split is creative against audience. If a brand-new ad in the same ad set starts strong, the people are fine and the old creative is worn. If new ads fade within days too, the audience has seen enough from you, whatever you show it, and widening it matters more than another round of creative.

What Meta's creative fatigue recommendation tells you

Meta also flags fatigue itself, as a recommendation. Its list of performance recommendations includes two about overexposed creative, "creative fatigue" and "creative limited", and Meta's developer documentation gives both the same explanation: cost per result for the ad set may be higher than ads you ran in the past, because its image or video has been shown to parts of your audience too many times (Meta). What the documentation does and doesn't tell you:

• No trigger is published. Meta gives no level of frequency, CTR or cost that sets the recommendation off.
• Applying it makes new variations for you. Accepting the creative fatigue recommendation uses generative AI to create new variations of your creative. For a supplement brand, read every variation before it runs: a line or image the AI adds is still your claim.
• Tools can get it as an alert. Software connected to Meta's Marketing API can receive a creative fatigue alert by webhook, rated low, medium or high, as in "Your ad fatigue changed from medium to high".

The alert also names the fixes Meta has in mind: a new ad with different images or videos in the same campaign, a wider audience, or Advantage+ creative. They match two of the outcomes above, a worn creative and a saturated audience. None of them fixes a broken checkout or a pricier auction, so run the questions first.

Read the decline in one ad's numbers

Once one ad looks fatigued, its columns show which part wore out. Here is the ad from the chart in weekly totals, read with the hook rate and hold rate conventions from our hook rate guide:

• Week 1: 60,000 impressions, 18,000 3-second plays, 3,600 ThruPlays and 780 link clicks.
• Week 3: 60,000 impressions, 12,600 3-second plays, 1,890 ThruPlays and 480 link clicks.

The calculator starts at week 3. Type in week 1 to see the gap, then try your own ad's first week against its latest:

Calculator

Hook rate and hold rate calculator

From the columns Meta reports for one video ad.

Impressions3-second video playsThruPlaysLink clicks

Hook rate

21.0%

3-second plays ÷ impressions

Hold rate

15.0%

ThruPlays ÷ 3-second plays

Link CTR

0.80%

link clicks ÷ impressions

Link clicks per 100 hooked viewers

3.8

Hook and hold rate are media-buyer conventions built from Meta’s columns, and teams define hold rate differently — write yours down and keep it. Both measure attention; judge the ad by the revenue per visitor it sends to your VSL.

Reset to the example

Week 1 works out to a 30.0% hook rate, a 20.0% hold rate, a 1.30% link CTR and 4.3 link clicks per 100 hooked viewers. Week 3 drops to 21.0%, 15.0%, 0.80% and 3.8. Link CTR is hook rate × clicks per hooked viewer, so the split shows where the 38% fall in CTR came from: hook rate fell 30%, clicks per hooked viewer 12%.

The opening wore out most. The likeliest reading is that people who have seen the ad recognize its first second and keep scrolling, so the first fix is new openings on the same body, the cheapest refresh there is. Hold rate fell by a quarter as well, though, so the body is wearing too: plan a new angle soon after the new hooks.

Before you conclude, check the placement breakdown for both weeks. If delivery moved toward placements where this ad has always earned fewer clicks, the averages fall without any fatigue at all, so compare each placement with itself.

Attention metrics stop at the click. Whether to keep spending depends on what the ad's visitors are worth on your page. With the rvs parameter on your ads, RevvyScales attributes each visit and sale to its campaign, ad set and ad (how rvs tracking works), so you can see whether a fading ad's revenue per visitor holds while its CTR slides. If it holds, the ad still sends good buyers, only fewer of them at a higher cost per click, and it can keep running on a smaller budget while its replacements prove themselves.

What to refresh first, and why a new color rarely works

A refresh works when the viewer meets something they haven't seen. Cosmetic changes rarely get you there: a new background color or a different music track leaves the same opening and the same promise, which is exactly what the audience has learned to scroll past. Change something the viewer will notice. Four refreshes for one tired ad, from the smallest change to the biggest:

Four ways to refresh one tired ad

A fictional greens powder whose winning ad is a 30-second demo of a two-minute morning routine

New hookNew formatNew presenterNew angle

New first three seconds, same body

Keep everything after the opening. Replace the first spoken line, the first image and the on-screen text.

Before: "My mornings used to start with coffee and nothing else."
After: a close-up of the scoop hitting the water, with the text "Two minutes. That's the whole routine."

The cheapest refresh. It fits when hook rate fell and clicks per hooked viewer held. Every new hook still sells the same promise.

The same angle, built differently

Rebuild the winning angle in a format this audience hasn't scrolled past yet.

• A static image with one line and the product
• A 9:16 selfie-style video for full-screen placements
• A split screen: the old morning routine and the new one
• Text on screen over the demo, for viewers with the sound off

It reaches people who skip that kind of ad. Statics are cheap to make, so they are a quick test of whether the angle still has life in it.

The same script, a different person

A customer who uses the product, a creator you pay or the founder, telling the same story in their own words.

The same 30-second structure, filmed in her own kitchen by a customer who drinks it before early shifts.

Real users only. The FTC says people can't talk about their experience with a product they haven't tried, and a paid relationship should be disclosed in the video itself, not only in the description (FTC). This isn't legal advice.

A different reason to buy

Same product, a different promise: from the routine to the taste, or to where the ingredients come from.

A taste test with the on-screen line "Greens that taste like lemon, not like a lawn."

The biggest reset of attention. It needs a new script and a proper test, because the angle is unproven. It fits when clicks per hooked viewer fell: the promise itself is worn.

Start where the numbers point. A falling hook rate with steady clicks per hooked viewer calls for new openings. Falling clicks per hooked viewer means the promise itself is worn, and a new hook on top of it won't change what's being promised; that's the moment for a new angle. To turn an angle into a tight 30-second ad, use the templates in our post on video ad scripts. And if you need new angles to choose from, start here:

How many new creatives you need, worked out from spend

Fatigue follows spend, so you can plan for it. Every winner carries a certain amount of spend before it fades, and your budget uses that up at a predictable speed:

Plan creative volume from spend

Winners used up per week=weekly spend ÷ spend a winner carries before it fades

New creatives to launch per week=winners used up per week ÷ hit rate

Weekly spend your creative supply can carry=new creatives per week × hit rate × spend per winner

Illustrative: $14,000 a week ÷ $18,000 per winner = 0.78 winners a week. With a 1-in-5 hit rate, 0.78 ÷ 0.2 = 3.9, so about four new creatives a week. A team that can make only two a week supports 2 × 0.2 × $18,000 = $7,200 a week.

Take an illustrative brand spending $2,000 a day, or $14,000 a week. Its last eight winners faded after a median of about $18,000 of spend each, where "faded" means the point at which an ad's three-day cost per purchase first stayed above the brand's target. One in five new creatives it tests becomes a winner. So it uses up 0.78 winners a week and has to launch about four new creatives every week just to stand still.

The plan also shows what changes the number. Double the budget and you double the creative you need, unless the audience grows with it. Iterations of a proven ad and brand-new concepts can succeed at different rates, so track the two hit rates separately and plan with both. And if your team can only make two creatives a week, the formula's last line is your real spending limit, whatever the ad account allows. Each new creative still has to earn its place, and the creative testing post sets how much it gets to spend before you judge it.

Why relaunching the same ad doesn't fix fatigue

The tempting shortcut is to duplicate the tired ad, or pause it and launch a copy, so it starts fresh. It doesn't, for a simple reason: fatigue lives in the audience, not in the ad's ID. The copy shows the same creative to the same people who stopped responding. The frequency column starts over on the new ad; the viewers' memory doesn't.

It also muddies your data. The ad's history is now split across two IDs, so its decline no longer reads as one line, sales attributed by ad are divided between the copies, and your test log stops matching what's running. When the copy fades within days, you've paid to learn what you already knew.

Three other mistakes to avoid:

• Raising the budget on a fading ad. More spend over the same audience adds impressions faster than reach, so frequency climbs faster.
• Killing a winner on one bad day. Daily purchase counts are small and jumpy. Judge three-day and seven-day trends against the ad's own first week.
• Tricks to get around review or delivery limits. Don't. They fix nothing the audience feels, and they put the ad account at risk.

The weekly fatigue review

Fatigue builds over days, so check for it on a fixed day each week, one ad at a time:

The weekly fatigue review

• For every ad that spent this week, compare its three-day CTR and cost per purchase with its own first week.
• Read frequency since launch next to them: rising frequency with falling CTR is the pattern to act on.
• Read Meta's recommendations, then confirm any fatigue flag against your own numbers.
• If every ad dropped at once, check the page, the offer, the checkout and purchase tracking before the creative.
• Launch one new ad into the same ad set to tell creative fatigue from a saturated audience.
• Compare each fading ad's revenue per visitor on your VSL with its earlier weeks.
• Lower budgets on fading ads that still beat your target; pause the ones that miss it, and never relaunch a copy.
• Brief next week's creatives from the plan: weekly spend ÷ spend per winner ÷ hit rate.
• Log the spend at which each winner faded, so your spend-per-winner figure stays current.

0 of 9 done