How to Find the Best Audiences and Use Them With Broad Match Keywords in Google Ads

Hi everyone! I’m Sasha, Digital Marketing Manager at Wrike. This is the story of how our team seeks out new approaches to search campaigns in Google Ads to improve the Conversion Rate in high-quality leads (HQL) and reduce cost per action (CPA). HQLs are leads with a high propensity to convert into paying customers. The higher the quality of your leads, the more likely they are to purchase your product or service. Thanks to our approach in the test period, we improved CR by 127% over the original version of the campaign.

This article will be useful to digital marketers who use audiences for observation and targeting in search campaigns and want to improve their performance.

What do you need to get started?

Each team contributes to the development and improvement of the product. Our goal is to find an approach for maximizing results with a minimum investment of time/money.

Since search campaigns are quite limited in terms of approach, we often combine some established approaches with opportunities provided by our advertising system. Today we’re going to talk about using Google Ads’ internal tool — Audience Manager.

Be warned straight away that for this to work as we expect it to, first-party data from the following sources should already be created in the account:

• Global site tag
• Google Analytics
• YouTube
• Google Play
• App analytics
• Customer data

If you are confident that you have such audiences, you can move on to the next step.

Google Audience Manager

This tool is available to all Google Ads advertisers and can be found by opening the “Tools and settings” section in the top panel and selecting the first item under the “Shared library” subsection:

Next, go to “Your data insights” and start working directly with the tool.

First, you will see the audience you can use by default, “All converters.” It consists of users who have converted by the conversions set at the account level. It is also possible to select the region where we want to find the audience — we chose the United States.

We can select any other audience available by clicking on the pencil. The same is available for selecting a region.

NB: It is important to remember that the larger the audience, the more accurate the results. Some audiences and regions will be impossible to select because they are too small for the tool to work correctly.

Segment distribution

When all the preparatory work is done, we can begin to explore the audiences that the tool suggests. Let’s go through all the proposed cuts in order.

• Demographics: This is quite useful information for finding audiences that are less likely to convert compared to the general audience.
• Locations: This information can be obtained from MCC/account-level reports. This shows the percentage of converted users from each region. A useful option is to see a cross-section by city and compare spending in these cities. Again, this data can be obtained by building reports in Google Ads.
• Devices: This tab gives us information on which users are prone to convert compared to all other users in the area. Be careful: The segment share will likely differ from the benchmark distribution if you exclude a particular device in your launched campaigns.

However, the decision to exclude DD, Location, or Device audiences is best made based on statistics rather than data from the tool.

Relevant segments (or where the magic begins)

Now the most interesting part, the recommended in-market and affinity segments. We see four columns with the name of the audience, index, the total size of the audience, and the number of campaigns in which that audience is used. Note that the list of audiences and the index of similarity will vary slightly for different regions.

In-market:

This will present Google’s internal audiences, graded by similarity index from larger to smaller. It is advisable to consider audiences with a higher index and look at those not used in any campaign. Since, in our case, we planned to use audiences as targeting, “targeting” outreach for us played a significant role.

Affinity:

The situation is the same with affinity audiences. We were surprised to see a fairly high index in users who belong, for example, to the category “Frequently Eats Dinner Out” or “Fashionistas.” However, it should be understood that the index is conditional. If you can look not at the similarity of account level conversion, but at the similarity of real customers, for example, the audiences can be quite different and their value is much higher.

Experiment (or how we were able to double our CR)

If you are actively using audiences on the campaign for observation or targeting, you will find that most of the audiences in those tables with a high index are already being used in campaigns (but not all of them).

For the experiment, we chose an original old campaign (RLSA type) that had accumulated statistics and used a combination of audiences and generic keywords in a broad match. For the experimental campaign, we added targeting audiences suggested by Google Audience Manager with the highest index (both in-market and affinity audiences) and audiences with the maximum amount of HQL at a low cost in the last year based on stats in the account and the same keywords in a broad match.

The experiment lasted almost four full weeks, using Google Ads’ internal “Experiments” function. It is worth noting that a few audiences remained unchanged in both the original and experimental campaigns because they fit the definition of ‘high-affinity index’ and ‘effective based on historical data,’ but these audiences accounted for less than 30% of the traffic in the experimental campaign.

The strategy chosen for both campaigns was basic — max conversions. The conversion set with offline HQL from CRM was the target conversion.

We increased the conversion rate in the HQL by 127% and reduced the cost of HQLs by 47%. Seems like a good result, don’t you think?