> For the complete documentation index, see [llms.txt](https://docs.adalysis.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.adalysis.com/meta-facebook/performance/ads/compare-groups-of-ads.md).

# Compare groups of ads

Put groups of ads side by side and see which one wins, and how sure that is

Compare formats, ad sets, campaigns, creatives, or your own themes, such as "UGC" against "studio", or two to six ads you tick. Adalysis adds up each group, picks the best one on the metric you choose, and tells you how confident it is that each other group is behind.

Groups are always formed from the ads the screen shows: the date range, the conversion action, and the filter all apply.

## Choose how to compare

![Ads, Compare menu](https://3203600314-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FGlXujWJyreRidNjZDCw2%2Fuploads%2FgJJF6A94joF3wycDbpeC%2FCA-01.png?alt=media)

1. **Compare**: opens the menu.
2. **Compare groups** opens the **Compare groups of ads** panel under the buttons, where you split the ads automatically or define your own groups. **Compare selected** compares the ads you've ticked in the grid, one ad per group. It needs between 2 and 6 ticked ads.

### Split automatically

![Ads, Compare groups of ads panel, automatic split](https://3203600314-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FGlXujWJyreRidNjZDCw2%2Fuploads%2FrZFpslPgEHadbZjV8Qis%2FCA-02.png?alt=media)

* ③ **Split automatically by**: choose it.
* ④ **The split**: pick what to split the ads by.
* ⑤ **Compare**: opens the results.

The splits:

* **Format**: image, video, and other ads.
* **Ad set** or **Campaign**: one group per ad set or campaign.
* **Image or video**: one group per image or video the ads share.

An automatic split makes up to 25 groups, largest spend first. With **Image or video**, groups are named Video 1, Image 1, and so on, and each shows a thumbnail. Ads that don't share a single image or video (carousels and posts, for example) go in a group called **Carousel or other**, which is never picked as the winner because it isn't one creative.

### Define your own groups

![Ads, Compare groups of ads panel, your own groups](https://3203600314-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FGlXujWJyreRidNjZDCw2%2Fuploads%2FO4lKjRQULvXVtEemmC4E%2FCA-03.png?alt=media)

* ③ **Define the groups**: choose it. Two groups are ready: **Group A** and **Group B**.
* ④ **Group name**: type your own, as "Image ads" here.
* ⑤ **Condition**: the same conditions as the filter's (for example **Ad name contains** "UGC"). **+ Add another** adds one more to the group.
* ⑥ **+ Add group**: adds a group. You can have up to 6.
* ⑦ **Compare**: opens the results.
* ⑧ **Delete group**: removes that group. It shows while there are more than two.
* A group's ads must meet all its conditions.
* An ad that meets more than one group's conditions counts in the first. The pop-up tells you how many ads did, and how many met no group.

## Read the results

![Compare groups of ads pop-up, side by side](https://3203600314-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FGlXujWJyreRidNjZDCw2%2Fuploads%2Fd7sEGjHTHKXhknm8ALwj%2FCA-04.png?alt=media)

1. **Conversion action selected**: what the conversion figures count.
2. **Confidence on**: the metric the winner is picked on.
3. **The winner's column**: shaded green.
4. **A group's label**: **Winner**, **Behind**, or **Not yet clear**. See below.
5. **Edit**: changes the minimums the confidence needs.
6. **Show ads** and **Filter the screen to it**: see [Act on a group](#act-on-a-group).

The pop-up opens with three lines at the top:

* **Comparison period**: the date range, and the period it's compared with.
* **Conversion action selected**: what the conversion figures count.
* **Split by**: how the groups were formed, and the filter in use.

Up to 6 groups are shown side by side, one column each, with every figure and its previous value. With more groups, they're shown as a ranked list on the chosen metric.

### Confidence on

**Confidence on** sets the metric the winner is picked on:

* **Cost per conversion** (the default; lower is better),
* **Conversion rate**,
* **Outbound CTR**,
* **ROAS**.

### Winner, Behind, and Not yet clear

Each group gets one label:

| Label                  | What it means                                                                                                                                      |
| ---------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Winner**             | The best group on the chosen metric. Its column is shaded green.                                                                                   |
| **Behind (95%+ sure)** | This group is behind the winner, and the gap is very unlikely to be luck. You can act on it.                                                       |
| **Behind (90%+ sure)** | This group is probably behind the winner, but there's still a small chance the gap is luck. Worth acting on with care, or waiting a little longer. |
| **Not yet clear**      | The gap could still be luck. Let the ads run longer before you decide.                                                                             |

"Sure" means statistically confident: Adalysis checks whether the difference is big enough, for the amount of data each group has, that it's unlikely to be chance. With many groups, it asks for a bigger difference, so a group isn't marked Behind just because one of many comparisons happened to look different.

A group needs enough data before it can win, so one lucky conversion never tops the list. See the minimums below. A group with no label didn't have enough data to be ranked, or there was no other group it could be compared with.

{% hint style="info" %}
Meta also counts conversions that didn't come from a click to the website, such as view-through conversions, lead forms, and messages. When a group has more conversions than outbound clicks, it has no conversion rate and isn't ranked on it. Hover its label to see why.
{% endhint %}

Groups often span ad sets with different audiences and budgets, so treat the comparison as a direction to explore rather than a strict test.

### Minimums for confidence calculations

The line under the results says the minimums used, for example "5 conversions. 100 outbound clicks."

![Minimums for confidence calculations pop-up](https://3203600314-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FGlXujWJyreRidNjZDCw2%2Fuploads%2FRAenXM5b2eageXV6vYKN%2FCA-05.png?alt=media)

1. **Edit**: opens the pop-up.
2. **Conversions**: the conversions each group needs to be ranked on cost per conversion, conversion rate or ROAS (default 5).
3. **Outbound clicks**: the outbound clicks each group needs to be ranked on outbound CTR (default 100).
4. **Save**: runs the comparison again with the new minimums. The minimums belong to the ad account and are shared by everyone who opens it.
5. **Reset to defaults**: puts back 5 and 100.

Each value is a whole number from 1 to 100,000.

## Act on a group

Under each group:

* **Show ads** lists the group's ads with their status and spend, and a link to open each in Ads Manager. A large group lists the ads that spent the most, up to 500. **Hide ads** closes the list.
* **Filter the screen to it** closes the pop-up and filters the Ads screen to that group's ads, counted on the same conversion action. It's shown when the group can be written as a filter.

**Download CSV** saves the comparison. **Close** closes the pop-up.
