Advertising
Meta ads: check this first before you change anything
· 2 min read · Oddbird
An ad that seems to be doing badly is often still learning or getting tired. Stepping in too fast makes it worse. These are the numbers to look at first, and in what order.
What you take away
- If an ad set is still in the learning phase, change as little as possible. A significant edit restarts the learning.
- Read from top to bottom: first cost per result, then clicks, then reach and frequency, and only then the creative.
- Change one thing at a time. Otherwise you will not know afterwards what made the difference.
The most common mistake with Meta ads is not a wrong audience or a weak text. It is stepping in at the wrong moment. After two days the numbers disappoint, so the budget goes down, the audience gets reworked and a new image goes in. Everything starts over, and you still do not know what was wrong.
First: is it still learning?
When an ad set has just started, Meta works out who your ad works best for. That is the learning phase, and Ads Manager literally says Learning next to it. Meta itself says about fifty results within a week are needed to get out of it. During that period costs fluctuate, and that is normal.
A significant edit, such as a much different budget, a new audience or a new ad, puts the set back into the learning phase. If you change something every few days, your ads never finish learning.
Then: the numbers, from top to bottom
| Metric | What it tells you | If it disappoints |
|---|---|---|
| Cost per result | What a lead, purchase or message costs you. This is what it is about. | Only look further if this stays too high after the learning phase. |
| Click-through rate (link) | How many people who see the ad also click through. | Low: the image or first line does not grab. High but few results: the problem is your landing page. |
| Cost per thousand impressions | What it costs to be seen. | High: the audience is small, or there is a lot of competition at that moment. |
| Frequency | How often the same person saw your ad on average. | If this rises while results fall, your ad is tired. |
That order matters. Fixing a low click-through rate is pointless if the real problem is your landing page, and an expensive audience is no problem as long as the cost per result is good.
You spot a tired creative by its frequency
An ad that did well at first and then slowly gets worse is usually not broken. The same people have simply seen it too often. You see it in a rising frequency and a falling click-through rate. The fix is then not a new audience but new creative: a different first second, a different face, a different angle on the same offer.
One thing at a time
- Wait until the learning phase is over, or until there are enough results to say something.
- Find the first metric in the table above that disappoints.
- Change only what belongs to that metric: the creative, the page or the audience.
- Give it a few days again, and compare with the period before.
Most ads do not fail because of a bad idea, but because of steering too often and too early.
This sounds slow, and it is a little. But it is the difference between guessing and learning. After a few rounds you know what works for your customers, and that is worth more than an ad that happened to have a good week.