Audience segmentation targeting

1. Click New group in the Audience segmentation section of your campaign properties.

2. Name your segment group and segments. Click New segment to add additional segments.

   Naming tip: It helps if your segment names match the ad or type of ads you plan to target to your segments. That makes it easier to remember which ads are targeted to your segments when you view your reports. Reports don’t indicate which segments your ads are targeted to.

3. Enter weights for each segment. See "Weights" below for more information.

4. Click New group again if you'd like to add more segment groups.

5. Save your campaign

6. After you save, Campaign Manager 360 segments all possible IDs based on your settings. See "About audience segmentation" below for more information.

1. In the Audience segmentation section of your ad properties, click Select segment and choose a segment. This segment will be the only group of users that can be served your ad.

   See "Targeting strategy" below for tips to set your segments up effectively.

2. On each placement in your campaign, include at least one ad that doesn’t require segmentation to serve. A non-segment ad is not required, but it's very useful. It allows you to serve something other than a default ad to users who don't belong to a segment. Here are the steps:

   1. Assign a standard ad to the placement. If this is the only non-segment ad, it should have no targeting criteria and use a creative that most browsers can handle. Otherwise, the non-segment ad may not serve to all or even most of your non-segment users.

   2. Set the priority of the non-segment ad as a higher number than the ads you target to segments. Enter the number in the ad properties.

      For example, if your segment ads have a priority of 5, set the non-segment ad to 6 or above. Do this to give your segment ads priority over the non-segment ad. You don't want your non-segment ad to have a chance of serving to users in segments. Users in segments should only get ads that are targeted to their segment.

3. Save your ad.

4. Make sure each placement in your campaign has at least one ad that isn't targeted to any segments. Since not every user can be segmented, you'll want to have an ad you can serve to unsegmented users. Otherwise, these users will only receive default ads. Your unsegmented ad should have a higher priority number than the segmented ads.

5. As your campaign progresses, you can compare the performance of ads targeted to different segments within your segment group.

   See the "Use case" and "Control/Exposed studies" sections below to learn more about comparing segments.

Campaign Manager 360 segments your audience based on cookies or device IDs. Each ID (cookie or mobile device ID) is assigned to a different segment. Once an ID is assigned to a segment, any user on a browser with that ID will be served ads you target to that segment.

Segmentation is effectively random: it's based solely on the digits in each ID, so it as nothing to do with any history of activity associated with the ID.

Here are two important concepts:

• Segment groups: Segments belong to segment groups. Each segment group divides all IDs across its segments. For example, if you create only two segments in a segment group, Campaign Manager 360 assigns all IDs to one segment or the other segment in that group.

  Every ID is automatically assigned to a segment in your segment group as soon as you create it. Those assignments last for the duration of the campaign, provided you don't change your segments.

  • You can have multiple segment groups. Each group segments all IDs. While your first segment group might divide all  IDs into two segments, your second might divide them into six segments. Each ID will belong to one segment in each group, because every group segments all your IDs independently of the rest. The ID might belong to the first segment in one group and the last segment in another group: segmentation is random.

• Weights: You can segment your audience evenly or in different proportions. That depends on the weights you set. See "Weights" for more information.

Weights determine how IDs are divided across segments in a segment group. Weighting is proportional. If one segment has a higher weight than the others in a group, it will have more IDs than the others.

For example, if two segments in a group are weighted 10 and 5, one segment will have twice as many IDs as the other (because 10 is twice as large as 5). Evenly weighted segments in a group have approximately even numbers of IDs.

Weighting only affects how segments in any given group are divided. The weight of a segment in one group does not affect segments in another group. Each group divides all IDs separately from the others.

All things being equal, ads you target to larger segments will receive more impressions than ads you target to smaller segments. That’s because the larger the segment relative to the other segments, the more IDs assigned to it relative to the other segments.

However, additional targeting criteria will narrow your audience for an ad even if the ad is assigned to a segment with a relatively high weight. For example, if you target an ad to a small geographic region, fewer users will be eligible to receive it. Even if the ad is targeted to a segment with a large portion of your audience, only some of the DoubleClick cookie IDs in that segment will be eligible to receive the ad.

To make sure your segmentation and weighting are the main factors in who sees your ads, it helps to use the same additional targeting criteria for every ad that's targeted to segments in your segment group. Or eliminate all targeting criteria except segmentation.

Say you want to compare the performance of two different kinds of ads: juice ads and soda ads. In this case, you might create a segment group in your campaign with two segments: a Juice segment and a Soda segment.

Next you target all your juice ads to the Juice segment and all your soda ads to the Soda segment. Now you can be sure that only users in the Juice segment will see juice ads, and that only users in the Soda segment will see soda ads. The segments don't overlap.

Next you assign a juice ad and a soda ad to the same placement.

• When a user in the Juice segment visits a site with this placement, Campaign Manager 360 serves a juice ad.

• When a user in the soda segment visits, Campaign Manager 360 serves a soda ad.

As your campaign progresses, you can check your reports to see which type of ad is performing better. Are there more clicks for users who only see the juice ads? Are there more Floodlight conversions generated by users who only see the soda ads? What are the differences that tell you which group of users is having a more effective ad experience?

Keep in mind that reporting doesn’t identify segments, so it’s up to you to check which ads are targeted with which segments.

Sample segment group

Let’s see how a segment group for this use case might look.

In this case, every ID is assigned to either the Juice segment or the Soda segment. The Juice segment is weighted 1 and the Soda segment is weighted 3. That means 3 times as many IDs are assigned to the Soda segment as to the Juice segment. A quarter of the users will be in the Juice segment and the other three quarters will be in the soda segment.

Next you assign a Soda segment ad and a Juice segment ad to the same placement. Since the Soda segment is 3 times larger, ads targeted to the Soda segment should get about 3 times more impressions than the Juice segment ads. That’s because 3 times as many IDs are assigned to the Soda segment, so there’s a better chance that a user visiting the site will have an ID assigned to that segment.

After your campaign is over, check your reports to see the performance of your juice ads and soda ads.

If some of your ads are experimental, you might set up a segment for just a small portion of your audience and target your experimental ads to it (an exposed segment). Then you can compare the performance of the exposed segment to that of your regular ads (a control segment). For example, if you wanted a ten times as many users in the control segment as the exposed segment, you would set a weight of 1 for the exposed segment and a weight of 10 for the control segment.

For an effective control/exposed study, create your segments before you start the campaign and leave them in place for the duration. The more impressions you have to work with, the more useful your data will be.

Improvements between a control segment and an exposed segment are called lift. Lift is the improvement from control to exposed. The improvement is measured as a percentage of the control segment's performance.

For example, if your control segment gets 20 impressions and your exposed segment gets 30 impressions, that’s a lift of 50%. Here's the formula for lift: (exposed metric - control metric) / control metric. In this case, that means: (30 - 20) / 20 = .5, or 50%.