Video Ad Fatigue: Definition, Thresholds and False Diagnoses

Video ad fatigue is the drop in an ad's performance when the same audience has been exposed too often to the same content: click-through rate and retention fall, while CPM and CPA rise. It is distinct from creative fatigue, which affects the asset itself. It is diagnosed from a cluster of signals, rising frequency and falling hook rate, never from frequency alone, and it is treated by renewing angles. On the accounts we audit, most cases labelled “fatigue” are actually something else. Here are the thresholds, the tests that settle the question, and the real cost of renewal. The starting point is already demanding: according to the 2025 edition of Kantar Media's TGI study, 47% of respondents say they never look at ads shown on social media.

Video ad fatigue: what is the exact definition in 2026?

Ad fatigue is an audience phenomenon: it is the performance degradation that appears when the same concept reaches the same people too often. The content has not changed. People's reaction has. It is the only definition that holds up operationally, because it points to a measurable cause, repeated exposure, rather than a vague impression that “the creative is worn out”.

French vocabulary systematically blurs two distinct things. English-language sources have separated them since 2025. According to AdLibrary, in its May 2026 overview, ad fatigue describes the decline at audience level, while creative fatigue refers to the cause at asset level: an edit, a hook or a promise that stops working even with people who have never seen it. This is not semantics. The two problems are not solved with the same budget.

Let's add a third thing that nobody names in French: account saturation. You have exhausted your addressable audience pool in a country, an age bracket, an interest. Here, the creative is not to blame. You can produce fifteen new videos, you will still be talking to the same 400,000 people. To give a sense of scale, Médiamétrie's L'Année Internet 2025, relayed by CB News, reports that 68% of French people check their social networks and messaging apps every day, or 43.9 million people.

Ad fatigue = people have seen it too much. Creative fatigue = the asset no longer works, even for a fresh audience. Account saturation = there is nobody new left to reach. Three diagnoses, three very different bills.

A second useful distinction: execution fatigue and concept fatigue. Execution fatigue is the same idea, poorly filmed or already seen fifteen times in the same edit; it is fixed with a variation at marginal cost. Concept fatigue is when the idea itself no longer says anything to the target, and you then have to change the way in. If the word “angle” is still fuzzy for you, first read the definition of a creative angle and its real cost: the rest of this article relies entirely on it.

The exposure context explains why the subject has become so hot in vertical video. Propulseur Digital noted in June 2026 that 78% of consumers come across ads on social media, 60% on websites and 56% on video streaming platforms. In a vertical feed, the ad takes up 100% of the screen and interrupts an entertainment session. Tolerance for repetition is structurally lower there than for a display banner that was never really looked at. According to Médiamétrie's L'Année TV / Vidéo 2025, French people spent 4h14 a day watching video content in 2025.

How does it show up in practice in a Meta or TikTok account?

The signals almost always arrive in the same order: hook rate degrades first, then CTR, then CPM, and CPA last. CPA is the signal everyone looks at. It is the latest and the most expensive to discover.

Adsights, in its July 2026 analysis, describes the same chain: a frequency above ~2.5 together with a falling share of “first-time” impressions, a CTR down 20% or more against its own baseline, a CPM rising for no identifiable bidding reason, and above all a hook rate that degrades before everything else on the video. These leading signals arrive one to two weeks before acquisition cost drifts.

Why the hook first? Because recognition is instant. Someone who has already seen your first second three times recognises it in 300 milliseconds and scrolls. They never even reach the message. Your 3-second retention drops while the conversion rate among people who do watch stays the same: this is the clearest signature of real audience fatigue. For the calculation method and platform-by-platform thresholds, see how to measure a hook, precisely.

The platforms' native labels, and their limits

Meta now displays “Creative Limited” and “Creative Fatigue” statuses directly in the manager. Atria noted in June 2026 that these labels are only usable at ad level: at campaign level, aggregation hides the problem. In practice, these alerts are a starting point, not a verdict. We see them trigger on ads that have just come out of the learning phase and that, two days later, become the best in the account again.

Our internal rule: no decision to cut or produce on the basis of a single indicator. You need two concordant signals over at least 5 days, read at ad level, and enough spend behind them (we come back to this below).

At what frequency can we really talk about fatigue?

There is no universal frequency, and a high frequency on its own proves nothing: it only becomes a signal if it coincides with a measured drop in CTR, hook rate or conversion rate. That is AdSpyder's position in 2026, and it is also what we observe on the accounts we have managed for three years.

That said, orders of magnitude do exist, and they converge. GoodMorning Co. published operational thresholds for DTC and e-commerce in May 2026: in prospecting, brief the replacement when frequency approaches 2.5 and deliver around 3.5; on broad top-of-funnel, monitor at 3.0 and refresh between 3.5 and 4.0, always subject to confirmation from the direction of CTR or CVR. Several practitioners place the conversion drop-off at around four exposures, though Meta has published no official threshold: treat it as a field benchmark, corresponding to a 7-day frequency of roughly 2.5 at the top of the funnel and 3.5 in the middle of the funnel.

GrowthJockey, cited in August 2025, places the onset of decline above 2.5 on most campaigns and speaks of immediate action above 3.0 in retargeting. On retargeting, we are more tolerant than these figures: a warm audience of 20,000 people mechanically reaches a frequency of 4 in ten days, and if ROAS holds, we touch nothing. The metric decides, not the threshold.

One research finding is worth knowing, because it contradicts the intuition that “the more you repeat, the more it sinks in”. Research on advertising frequency shows diminishing returns beyond a few exposures: the marginal effect of each new impression falls quickly, without a precise figure being attributable to a single source. Put bluntly: beyond a certain rank, each additional impression costs you money and brings in less than the first.

Still, frequency is distorted by CPM. A CPM that climbs 40% in November makes you buy fewer impressions for the same budget, so it lowers your frequency while degrading your CPA. You conclude “no fatigue” when it is the auction that is moving. To calibrate your CPM, CPC and ROAS benchmarks against the French market, use the 2026 French CPM, CPC and ROAS benchmarks before interpreting your own curves.

And if it isn't fatigue? The 4 costliest false diagnoses

The most frequent and most expensive false diagnosis is confusing fatigue with under-delivery: according to Succession Media in 2026, 80% of creatives never reach 100,000 impressions; most ads die from a lack of budget, not wear. We have seen brands order six replacement videos for assets that had spent €60 each. That is money thrown away twice: the production was pointless, and the creative that might have won was killed before it existed.

Second trap: the learning phase restarted. Meta for Developers states that an ad set needs about 50 optimisation events over 7 days to stabilise, and that a significant edit can reset the counter to zero. Did you change the budget by 30%, add a placement, edit the copy? Your rising CPA is not fatigue, it is a restart. Waiting costs zero euros. Producing costs the price of a shoot.

Third trap, the aggregated reading. A single ad failing in an ad set of six drags the whole average. You see “the account is fatigued”, when what you need is to cut one asset. Fourth, bidding seasonality: a CPM rising because everyone is buying the same inventory on Black Friday has nothing to do with your creative.

Here is the grid we use in audits. It reads line by line, and it has already saved several thousand euros of unnecessary production for clients who called us to order a refresh.

On line 3, a budget precision that avoids 90% of mistakes: Succession Media recommends allocating 2 to 3 times the target CPA to each test, with a floor of around $100 to $150, before passing any judgement. Below that, you are measuring nothing, you are looking at noise. It is the direct consequence of the media floor rule, which we have detailed elsewhere on this blog.

How long does a vertical video actually live before it fatigues?

Cycles have shortened: according to AdLibrary in May 2026, with Meta's Andromeda ranking system weighting creative signals more heavily, a concept that used to last six weeks now burns out in two to three. Your creative hasn't got worse. The machine sorts faster and replaces faster.

Archive data give another order of magnitude, and need to be read with care. AdSpyder, on a sample of archives from May 2026, observes a median run time of 4 days on Meta, 5.1 days on YouTube and 43.3 days on LinkedIn.

Beware of the misreading. A median of 4 days does not mean a Meta video fatigues in 4 days: it means half of the ads are stopped before that. The two populations are mixed, tests killed within 48 hours and evergreens running for six months. This is exactly the bias created by line 3 of the previous table.

What we observe on our own e-commerce accounts in France, in 2025 and 2026: a creative that has survived its first 10 days and runs on a significant budget generally lasts between 3 and 6 weeks before its hook rate drops. The rare winners exceed three months, often because they have been re-hooked two or three times. This is an internal observation, on a sample of DTC and retail clients, not published data; treat it as a benchmark, not a law.

What we cannot measure cleanly, and it should be said: isolating audience fatigue from bid drift in peak season. In November, the two move at the same time and no mainstream tool separates them. We then work by comparison with a control creative launched the same week.

How much does ad fatigue cost? The monthly calculation in euros

Fatigue is not a creative problem, it is a recurring budget line: monthly renewal cost = (30 ÷ useful lifespan in days) × number of active angles × your unit cost per video. As long as this calculation hasn't been done, you suffer fatigue instead of funding it.

Three variables, each measurable in your account. Useful lifespan: take your last five winning creatives, count the days between peak performance and the confirmed drop in hook rate, and take the median. Number of active angles: how many different ways in are running at the same time (price, problem, social proof, demonstration…). Unit cost: what you actually pay for a video ready to run, rights included.

Multiply the last column by your real unit cost and you get your fatigue bill. For the scale-up with €30,000 of media, 15 videos a month: if your unit cost is €250, that is €3,750 of production, or 12.5% of media. If your unit cost is €900, it is €13,500, or 45%, and at that level the production model is the problem, not fatigue. This is precisely when you need to choose between one-off production and a subscription: we detailed that reasoning in translating this pace into monthly throughput.

The good news is that not every renewed unit costs the same. Re-hooking an existing asset costs a fraction of a shoot. That is the subject of the next section.

Which renewal protocol should you adopt: rotation, new angle or new concept?

The level of response must match the level of the problem: an audience signal is treated first with media or a rotation, an asset signal with an iteration, and only concept fatigue justifies a new shoot. Reversing this order means paying full price for a problem that costs zero euros.

AdStellar mentions two complementary countermeasures: audience expansion, which exposes the creative to fresh users to buy it time, and creative rotation, which shows several variations to the same audience. The first is free in production terms. The second requires having stock.

The four levels, from cheapest to most expensive

• Level 0, media. Broaden the audience, open a new country, change exclusions. Production cost: zero. To be tried systematically first when the first-time impression ratio drops.
• Level 1, new hook on an existing asset. First three seconds replaced, body of the video kept. This is the response to the signal of a collapsing hook rate with CTR still stable. Typical turnaround: 24 to 48 hours in-house.
• Level 2, iteration. Same angle, different execution: another face, another setting, another editing rhythm, another call to action. This is what you really buy in 70% of “refresh” orders; see what you actually buy when you order an iteration.
• Level 3, new angle, new shoot. Another psychological way in, another promise. Full cost. We go there when two consecutive iterations on the same angle haven't restored the hook rate.

A firm position, which doesn't suit us commercially: if you call us for a refresh and we see that your creatives plateau at 30,000 impressions, we will tell you to put budget back before ordering anything from us. Producing six videos for an under-delivered account means manufacturing six more failures and burning trust in the video lever.

The format itself can extend lifespan. Amazon Ads reports for Sony (LinkBuds S) a cost per add-to-cart 77% below its Amazon average and an increase in brand searches. The mechanism is simple: a format that asks for an action isn't consumed passively, so it is recognised less quickly.

What questions should you ask your agency to find out whether it can manage fatigue?

An agency that manages fatigue can give you the median lifespan of the creatives it has delivered to you, and the number of angles it holds in reserve for you. If it can't answer these two questions, it sells videos, not performance.

• What is the median lifespan, in days, of the last five creatives you delivered to me that actually spent budget?
• How many angles have you tested on my account, and how many remain in reserve, unproduced?
• What is your turnaround between detecting a hook rate drop and putting a new hook live?
• On which indicator do you decide to cut a creative, and from what minimum spend?
• Do you deliver several hooks per shoot, or a single edited version?

This last question is the most discriminating. A studio that has understood fatigue always shoots two to four alternative openings during the session, because it knows they will be used in fifteen days. A provider that delivers a single file will sell you a new shoot every time performance drops.

One last point, which applies to the whole subject: video ad fatigue is not an anomaly to correct, it is the normal state of an account that spends. Brands that last have stopped thinking in campaigns and think in terms of a stock of available angles. They know how many days a creative lives with them, how many they must add each month, and what it costs them to the euro. The others discover the problem on a Monday morning, when CPA has already doubled, and then pay for renewal in a rush, at the worst moment and the worst rate.

Frequently asked questions

What is the difference between ad fatigue and creative fatigue?

Ad fatigue sits at audience level: the same people have seen the ad too often and stop reacting to it. Creative fatigue sits at asset level: the hook, the edit or the promise no longer work, even with an audience discovering the video. The first is often treated with media, the second requires a new production.

At what frequency should you renew a video creative?

According to GoodMorning Co. in May 2026, you brief the replacement at a frequency of around 2.5 in e-commerce prospecting and deliver around 3.5; on broad top-of-funnel, monitor at 3.0 and refresh between 3.5 and 4.0. These thresholds are only valid when confirmed by a drop in CTR or conversion rate. Frequency alone never proves fatigue.

How do I know whether my ad is fatiguing or simply lacks budget?

Look at cumulative impressions and spend. Succession Media stated in 2026 that 80% of creatives never reach 100,000 impressions: most die for lack of delivery. Until an ad has spent 2 to 3 times your target CPA, with a floor of around a hundred euros, you are measuring noise, not fatigue.

How long does a vertical video ad live in 2026?

AdLibrary noted in May 2026 that with Meta's Andromeda ranking, a concept that used to last six weeks now burns out in two to three. AdSpyder archives show a median run time of 4 days on Meta and 43.3 days on LinkedIn, but these medians mix tests cut early with evergreens: they are not fatigue points.

What is the earliest signal of video ad fatigue?

Hook rate, that is, 3-second retention. Adsights noted in July 2026 that it degrades before CTR and CPA, one to two weeks earlier. A hook rate that falls while viewers' conversion rate stays stable signals that the audience recognises the creative, hence real exposure fatigue.

What does creative renewal cost each month?

Apply the formula: (30 ÷ useful lifespan in days) × number of active angles × your unit cost per video. An account with €10,000 of media, 3 active angles and a useful lifespan of 14 days needs 6 to 7 videos a month. The production line then generally represents between 10 and 25% of the media budget.