Creative fatigue or audience saturation? A ten-minute test

This one is a test, not a story. Ten minutes, three columns of data you already have, and a decision at the end of it.

The problem it solves shows up in every market I have worked in, usually in the second week after a peak season. Click-through rate is down. Cost per result is up. Somebody says "the creative is tired" and a brief goes out for new ads. About half the time that is right. The other half, the creative was fine and the audience was used up. New ads go live, nothing improves, and two more weeks of budget are gone before anyone looks again.

The two problems have the same symptom and opposite fixes. Fatigue means the same people have seen the same ad too often. Saturation means you have already reached nearly everyone in the audience you defined, so the platform is showing the ad to the same people because nobody new is left. Fix fatigue by changing the creative and leaving the audience alone. Fix saturation by widening the audience and leaving the creative alone. Do the wrong one and you pay for a fix to a problem you do not have.

The test

Pull three numbers by week for the last six to eight weeks, per ad group: CTR, frequency and reach. Put them on one chart. The shape gives you the answer.

Saturation looks like this: reach climbs fast, then goes flat. Frequency keeps climbing because there is nobody new to show the ad to. CTR falls in step with frequency.

Fatigue looks like this: reach keeps growing, week after week. Frequency stays roughly flat, say two to three. CTR still falls. The audience is fresh and still they do not click. The ad has stopped working, often because a competitor is running something better or the offer no longer fits the season.

There is a third shape worth knowing. Reach flat, frequency flat, CTR falling. That is usually neither. Look at the auction. Your rank has dropped because someone else is bidding harder, and you are now winning cheaper, worse placements.

The decision table

1. CTR down, frequency up, reach flat. Saturation. Widen the audience: add lookalikes, loosen the interest filter, open a new region. Do not touch the creative.

2. CTR down, frequency flat, reach growing. Fatigue. Rotate the creative: new hook, new first three seconds, new offer. Do not touch the audience.

3. CTR down, frequency flat, reach flat. Rank. Check impression share lost to rank and your average position. Fix the bid, the relevance or the landing page. Leave the audience and the creative alone.

4. CTR down, conversion rate up. Leave everything alone. Fewer, better clicks is often exactly what you asked the algorithm for.

Where to find the inputs: Meta reports first-time impression ratio and frequency directly. Google Ads gives you "search lost impression share (rank)" and frequency for video. DV360 has a frequency distribution report. Most DSPs have a reach curve if you ask. None of this needs a data scientist.

The "just add budget" trap

The related mistake is adding budget when the problem is rank. A team sees delivery falling and raises the daily budget. Nothing changes, because the campaign was never budget-limited. It was losing auctions.

Three lines of arithmetic settle it. Divide yesterday's spend by the daily budget. If the answer is under 0.9, budget is not the limit. Now look at impression share lost to rank. If it is above 20%, rank is the limit. Adding budget to a rank problem only raises the price you pay for the same losses.

Why a system should run this, not a person

None of this is hard. It goes wrong because it needs three numbers from two or three platform screens, looked at as a shape rather than a value, every week, on every ad group. Humans do that for the top five campaigns and hope the rest are fine. I did exactly that for years.

At AdNexa the platform runs this check daily on every ad group across the 15 platforms we connect to. When it flags fatigue or saturation, the recommendation names the exact setting on the exact platform: which audience to widen, or which creative to retire. Then it remembers. Twenty-four months of account history means that when the same audience saturates at the same frequency next season, the system says so before the CTR falls. Most tools start every morning with no memory of yesterday. A media team that has to rediscover the same lesson every season is paying for it every season.