Now open to non-executive and advisory board roles. to board roles.Boards & advisory
  1. Home
  2. Insights
  3. Meta ads
Meta ads

How to Tell an Ad Is Finished

An ad is finished when click-through rate has fallen from its own best fortnight, cost per result is rising while CPM holds steady, and the comments have gone quiet or turned. Require two of those three before acting, and then change the hook alone so the result stays readable.

By Stephen EllulPublished 1 Oct 20267 min read
How to Tell an Ad Is Finished cover

I have killed a lot of ads too early and kept a few far too long, and both mistakes came from the same place: I was watching frequency. It is the number everyone reaches for when an ad starts feeling tired, and on its own it is close to useless.

Here is the problem with it. In a market the size of Malta, frequency climbs permanently because the audience is finite, so a perfectly healthy ad will sit at a frequency that would alarm anyone used to larger markets. Meanwhile a fresh audience can get bored of a weak hook within days, at a frequency of barely two. The number moves for reasons that have nothing to do with whether the ad still works.

So I stopped using it as a verdict and started using a short set of signals that compare the ad against its own history rather than against a benchmark somebody else published.

How do you know when an ad is finished?

You compare the ad against its own baseline, not against a benchmark. An ad is finished when click-through rate has fallen meaningfully from its own best fortnight, cost per result is climbing while CPM holds steady, and the comments have gone quiet or turned. Any one of those on its own is noise.

That last clause matters more than the rest. I have watched too many people kill a winner on one bad week. Weekly numbers on a small budget swing for reasons that are nobody’s fault, so I want two of the three signals pointing the same way across a fortnight before I act.

Why is frequency a bad signal on its own?

Because frequency measures exposure, not fatigue. It tells you how many times the average person saw the ad, and nothing about whether seeing it again changed their mind. A small market produces high frequency on every campaign that runs for any length of time, and a bad hook can exhaust a brand new audience at a frequency under two.

I still look at it, but only as context for something else. If cost per result is rising and frequency is flat, something outside the creative has changed, usually the audience or the offer. If cost per result is rising while frequency climbs hard, the creative explanation is at least plausible.

What does a falling click-through rate actually tell you?

That fewer people find the opening interesting than did a fortnight ago. Click-through rate is the cheapest early signal you have, because it moves before cost per result does and it needs far less volume to read. The useful comparison is always the ad against its own best period, never against an industry average.

Set the baseline deliberately. I take the ad’s best complete fortnight after the learning phase and treat that as its ceiling. A drift of a few percent is nothing. A quarter off the ceiling, sustained, is the ad telling you the opening has been heard.

Why does a stable CPM with a rising cost per result matter?

Because it isolates the problem. If CPM is flat and your cost per result is climbing, you are paying the same to reach people and getting less back, which points squarely at the message. If CPM is climbing too, you may simply be in a more expensive auction, and the creative might be innocent.

That one distinction has saved me more money than any other check on this list. Auction pressure around the holidays, a competitor arriving, a seasonal swing: all of those raise CPM and look exactly like creative fatigue on a dashboard that only shows cost per lead.

Can comments tell you anything useful?

Yes, and they are the most underrated signal in the account. Engagement falling silent usually means the ad has stopped provoking a reaction rather than stopped being seen. Comments turning sceptical or repetitive, with the same objection appearing again and again, means the argument itself has been answered in the market and needs replacing.

I read the comments before I read the numbers on anything that has been live more than a month. They tell you which part is tired. A quiet comment section says the hook has gone invisible. The same objection five times says the claim has lost its argument, which is a different repair.

What do you change first?

The hook, almost always, and only the hook. Change one thing so the result is readable: the opening claim, with the audience, the offer, the budget and the format left exactly where they were. If you refresh the footage and the targeting and the landing page together, you have learned nothing about any of them.

What I am not going to do here is relitigate why the hook is the thing that wears out rather than the visual, because my agency has already written the long version of that argument with the account data behind it. If you want the reasoning, read why your best ad stops working. This piece is about the diagnosis, that one is about the cause.

What should you do before you touch anything?

Write down the baseline. Record the ad’s best fortnight for click-through rate, cost per result and CPM, note the date, and keep it somewhere you will find it again. Without that, every future judgement about whether the ad is finished becomes a memory test, and memory flatters whatever you did most recently.

A short routine I run on every account:

  • Compare against the ad’s own best fortnight, never an industry benchmark
  • Require two of the three signals before acting: click-through decline, rising cost per result with flat CPM, comment sentiment
  • Read the comments on anything live more than a month, before opening the numbers
  • Change the hook alone, and give it a fortnight to read
  • Keep the retired ad. Decayed arguments often work again months later on a partly renewed audience

That last point surprises people. An argument that has been heard is not an argument that is wrong, and in a small market the population of people who have not yet considered it refills slowly but genuinely. I bring old winners back more often than I write new ones.

None of this needs a dashboard or a tool. It needs a baseline written down and the discipline to change one thing at a time, which is harder than it sounds when spend is rising and someone wants an answer today. If you would rather have a team running that discipline on your account every week, that is what my agency does.

Frequently Asked Questions

Is there a frequency number at which I should refresh an ad?

No, and chasing one will cost you money in a small market. Frequency climbs permanently on a finite audience, so a healthy campaign in Malta can sit at a level that would look alarming elsewhere, while a weak hook can exhaust a brand new audience below two. Use frequency as context for a cost change, never as the trigger on its own.

How long should I wait before deciding an ad is finished?

A fortnight of consistent movement, compared against the ad’s own best fortnight after the learning phase. On a small budget, weekly figures swing for reasons nobody controls, so a single bad week is not evidence. If two of the three signals point the same way across two weeks, that is a decision rather than a reaction.

Does creative fatigue mean I need new video every month?

Usually not. The thing that wears out is the argument in the opening, not the footage, so new production around the same claim tends to buy very little. Rewriting the hook over existing assets is faster, cheaper and more readable as a test. Reserve new production for when you have a genuinely new argument to make.

Can a decayed ad work again later?

Yes, more often than people expect. An argument that has been heard is not an argument that is wrong, and the pool of people who have not yet considered it refills over months even in a small market. Keep retired winners rather than deleting them, and reintroduce them after a gap instead of writing something new from scratch.

What if cost per result is rising and CPM is rising too?

Then suspect the auction before the creative. A seasonal spike, a new competitor bidding or a holiday period will raise CPM and look identical to fatigue on a report that only shows cost per lead. Check CPM against the same ad a month earlier, and if both moved together, your creative may be innocent.

Read next

Meta ads

Facebook Ads for Restaurants: A Complete Guide

How to run Meta Ads for your restaurant: campaign objectives, local targeting, creative that converts, and how to measure results properly.

27 Oct 2026 · 5 min read
Meta ads

What Is a Good Cost Per Lead on Facebook Ads?

Industry CPL benchmarks for Facebook Ads in 2026, plus the exact formula to calculate your own target cost per lead from your business unit economics.

27 Oct 2026 · 6 min read
Meta ads

What Is a Good ROAS for Meta Ads? 2026 Benchmarks by Industry

A good ROAS for Meta Ads is typically 3x–5x for e-commerce, but varies significantly by industry and margin. Here's the full breakdown.

21 Aug 2026 · 5 min read

Book 30 minutes.

I'll look at where your leads come from today and tell you straight whether I can help.

Rather build it yourself? The 30-Day Pipeline™ walks you through it in your own accounts.

The Meta ads checklist

The checks I run on every new account before spending a euro.

Check your inbox. The checklist is on its way.