Methods for A/B testing display ads go beyond the obvious with just testing creative. Sometimes we want to test different audiences to ensure we are attracting the right consumers at the right stage of the conversion cycle. This is particularly important when we have multiple consumer groups that have shown interest in your offering, and you want to determine how best to balance ad spend with return on investment.

This way, we’re getting the most out of that spend, with the goal of driving revenue. We test to figure out who we should be going after, then we can figure out which targeting practice is best.

There are several types of targeting that should be considered. When A/B testing targeting audiences, we can A/B test within one target category or two targeting categories against one another. For instance, we might test two audiences with different age demographics to see if an ad performs better with one or another.

Below, I’ve included a basic overview of each first type of targeting as a refresher with their associated advantages and disadvantages. For each, I’ll give an example of an A/B testing use case:

Demographic: Demographic targeting is based on consumer attributes, such as age, gender, household income, etc. Demographic targeting is extremely easy to scale because there are a wide range of 3rd-party data providers. However, given the nature of 3rd-party data, the resulting lists won’t be nearly as accurate as 1st-party data. We typically use demographic targeting for wide-reaching branding and awareness campaigns that aren’t tied to CPA.

Example: For a product like the Amazon Echo that appeals to a large demographic group, we could target fathers age 60+ vs. professional women in the 20s-30s to see which shows better potential for conversion.

Behavioral: Behavioral targeting identifies an audience based on specific online activity. An example would be someone searching for sneakers online. A shoe retailer can then use that cookie data to reach a user with an ad for shoes. Like demographic targeting, behavioral data is a type of 3rd-party data targeting and thus comes with the same advantages of being scalable; it’s also more accurate because we’re targeting based on a specific activity.

Example: For an airline, we might want to A/B test ads on users who regularly read travel blogs vs. users who are browsing hotel sites.

Lookalike: With lookalike targeting, we take a segment of people, such as our current customers, and target people with similar characteristics. An advantage of lookalike targeting is that if we know what resonates with a group of target customers, we can easily extend the reach to more customers who may contain a similar mindset, respond to the same types of messaging, etc.

Example: A B2B company might A/B test lookalike audiences based off of current MQLs vs whitepaper downloads.

Contextual: Contextual targeting is different than the other types of targeting we’ve covered so far because we aren’t targeting based on user characteristics or actions. With contextual targeting, we are reaching users who visit a specific site. Those sites are selected based on the category/industry or keywords associated with that site. The advantage of contextual, if done correctly, is that we will hit users who are in the right mindset and are interested in what you are offering. For instance, an airline might target vacation and destination sites because users researching places to travel are likely in the right mindset to purchase airfare.

Example: Is a golf club company better off targeting customers on the PGA website as opposed to golf course websites?

Retargeting: Retargeting allows you to target users who have visited your site. These users came to your site with interest and may have completed certain actions. Retargeting is a way to re-engage them so they move along in the funnel, or encourage them to return for a repeat purchase. Compared to other types of targeting, we tend to see much higher response and CTRs because users are already familiar with the site. The audience size we are reaching is typically much smaller with retargeting, and it’s more expensive to reach, but on the flip side, retargeting has higher conversion rates and demonstrates relatively higher ROI.

Example: With retargeting, because we are targeting users who have been to your site, we wouldn’t likely A/B test within retargeting, but we could test against another targeting group to see what’s a more effective use of spend.

Search retargeting: With search retargeting, we aim to engage users who have shown intent, but instead of targeting website visitors, we are targeting based on user search history.

Example: As seen below, we could A/B test on users looking for tips for sleeping better vs. users searching for a new bed.

Of course, as mentioned above, we can also test different targeting methods against each other. For our Amazon Echo example, we can test 60+ year old dads against people searching for 2-way wireless speakers. As marketers, to stay ahead of the curve we have to get creative, and A/B testing targeting tactics is an important way to do that.

For help on how to optimize your A/B testing efforts, contact the 3Q Display team today!

Leave a Comment

Luis Valles
Luis joined 3Q Digital in September 2016. He graduated from the University of Texas in 2014 with a B.S. in Advertising. He has held positions at The Richards Group and GSD&M, working on accounts such as GameStop, Fram Oil, and Air Force. When he is not working, Luis likes staying active by playing basketball, bouldering, mountain biking, and running with his dogs.