Many advertisers are well aware that Google has targeting options available in AdWords that give brands the ability to adjust bids, ad copy and landing pages for demographic groups based on age, gender and average household income.
These levers can be useful for targeting search campaigns only to the most relevant groups, as well as for improving the performance of particular demographic groups by creating catered experiences for them.
However, the current ways in which age, gender and average household income work together make it difficult to create granular demographic targets at scale. Here, we’ll illustrate the troubles that can arise, as well as explain why there is growing need for demographic targets that “talk to one another.”
The current situation with age and gender targeting
Currently, targeting an AdWords campaign to a particular age group or gender is done by exclusion, i.e., an advertiser adds age and gender groups to a campaign and then assigns those groups that should not be targeted as exclusions.
So, if a campaign should only target males between the ages of 18 and 24, all other ages and gender targets, including “Unknown” (which includes members of all ages and genders), should be added to the campaign as exclusions.
In the early goings of Google’s testing for age and gender targeting, advertisers had the option to select “Target and bid,” as well as “Bid only,” as shown by the screen shot below taken in early 2015.
The former limits targeting solely to those ages and genders launched as targets, while the latter simply layers the target on for bid adjustments and does not restrict a campaign’s targeting solely to those groups added.
However, it’s only by audience exclusion that age and gender targets can now be used to narrow targeting down, as all age and gender targets are set to “Bid only.” Both “Bid only” and “Target and bid” are still available for other audience targeting options, such as Remarketing Lists for Search Ads (RLSA) and Customer Match.
Unfortunately, age and gender can’t be combined to target a specific slice of the population across both attributes. That is to say, there’s no way to create a single target for males between the ages of 18 and 24 — advertisers have to exclude all of the ages and all of the genders that don’t match the targeted group separately. This makes it difficult to target a particular group in a separate campaign while also continuing to target all other genders and ages in a different campaign.
In the example of wanting to target males between the ages of 18 and 24, a brand could duplicate an existing campaign and exclude all ages outside of the 18-24 range (including “unknown”), as well as all gender targets not related to males (including “unknown”) in the new campaign.
In order to ensure that all of the individuals in that audience go to the new campaign, the original version of the campaign needs to be updated to exclude 18- to 24-year-old males. If the new campaign has higher bids than the original campaign, then 18- to 24-year-old male traffic should tend to correctly head to the new campaign, but it’s far from certain without proper exclusion in the original campaign.
However, if the original campaign is set to exclude males and people 18-24, all males would be excluded as well as all 18- to 24-year-old people. There’s no way to exclude only 18- to 24-year-old males.
That means that in order to cleanly segment traffic, as well as continue serving ads to all ages and genders, the brand must triplicate (it’s totally a word) the campaign and use the third version to fill in the gaps created by excluding segments from the original campaign.
So, if the original campaign were set to exclude all males, the third campaign would target all males not aged 18-24. Thus, one copy of the campaign would target males 18-24 years old, one would target all the other males, and the original campaign would target all non-males.
Imagine if the brand also wanted to incorporate average household income and wanted to target the top 10 percent highest income 18- to 24-year-old males in a separate campaign. They might have to create four different versions of the campaign to cover all the gaps and not overlap coverage!
Note: Age and Gender targets can be set at the ad group level, but for the sake of simplicity, in this post I refer only to campaign settings. Some duplication efforts can be carried out at the ad group level within a campaign. Average Household Income (HHI) targets can only be set at the campaign level.
Having to create multiple campaigns for the purpose of targeting a specific audience segment is annoying, and it’s a real impediment to granular targeting of different groups using age, gender and household income targets.
If adjusting bids for an audience segment is all that’s necessary, an advertiser could just layer on age and gender bid adjustments to existing campaigns and never duplicate anything, but there are big limitations to this strategy as well.
Dangers of a bid modifier stack attack
One well-known problem that Enhanced Campaigns have had since their inception is that it can be difficult to create effective bid modifiers because of how they stack on top of one another in adjusting bids. This is a problem because not all bid modifiers are independent variables, such as age and income. That means that advertisers might want to apply one modifier calling on multiple variables.
For example, say you want to push bids 25 percent for 18- to 24-year-old males searching from an area with a top 10 percent average household income without duplicating a campaign.
Since bid modifiers stack on top of one another, you’d either have to pick just one of the three attributes of age, gender and HHI to bid up 25 percent relative to the rest of the population, or set the modifiers of each of the three criteria such that they add up to a 25 percent push. In either case, bids are going to be adjusted for more people than just the individuals within the targeted group of high income, young males.
This strategy also wouldn’t allow a brand to adjust ad copy or landing pages for the target group without adjusting them for all demographic groups targeted by the campaign, and the only way to adjust copy and landing pages for just a targeted age-gender-income combination is to create separate campaigns.
Conclusion
The current AdWords levers advertisers have at their disposal to target users based on attributes like age and gender are nice, and they are certainly, at the very least, useful for collecting data on what types of users are searching and clicking on ads.
However, in practice, these targets are pretty unwieldy when trying to target specific groups of individuals based on multiple attributes.
There’s been nothing to suggest Google plans to abandon audience stacking for the purposes of calculating bids, or to allow advertisers to target a campaign to a particular slice of the population calling on multiple variables.
In the big scheme of things, these issues aren’t meaningfully hampering advertisers from investing in Google at the moment, but they do present a challenge to fully embracing audience-based optimizations. Google updates over the last couple of years, from the release of Customer Match to the expansion of maximum membership duration for RLSA audiences, have made it easier for advertisers to target members of known groups. Hopefully, it will continue to embrace updates that give advertisers better control in wielding audience targets.
Contributing authors are invited to create content for Search Engine Land and are chosen for their expertise and contribution to the search community. Our contributors work under the oversight of the editorial staff and contributions are checked for quality and relevance to our readers. The opinions they express are their own.
