Digital marketers considering or utilizing Gmail campaigns in Google Ads have a good number of audience types to choose from. In an ideal world, one would create multiple audiences and test each audience against each other. However, as many advertisers are well-too-aware, there are frequently instances where ideal circumstances do not exist for true experimentation. Whether because budget is limited or because the Gmail campaign will be relatively short term, there are many situations where an advertiser will have to privilege one or more audience type over others.

The results discussed in this blog post may be of use to advertisers who find themselves in that situation. On behalf of a lead-generation client in the real estate industry, we’ve run Gmail campaigns for 20 of their properties, generating a sizable data set over the past 120 days. These campaigns have been segmented at the ad group level by audience type, which gives us an opportunity to evaluate the performance of these different audience types across a variety of circumstances.

Gmail Audience Types: A Brief Primer

Prior to discussing the results of the case study, let’s dive into an explanation of each audience type that accumulated data during the examined period:

  • Remarketing – as would be the case for general display remarketing campaigns, this audience type targets users who have visited the client’s website. In this case, since repeat conversions are relatively rare and don’t produce much value for the client, we’ve excluded past converters from those audiences.
  • Keywords – Google analyzes a user’s Google and Youtube search history, and if that search history matches selected keywords, serves them the associated Gmail ads. Our keyword audiences targeted keywords that we would expect to match with users searching for apartment rentals.
  • In-Market – For this audience type, Google will serve ads to users who have been algorithmically determined to be in-market for a product or service. In this case we targeted users who were determined to be in-market for “Apartment Rentals”.
  • Affinity – Affinity audiences target users based on their habits and interests. We select proxy audiences for which users in those groups would be likely to be in in the client’s targeted demographic (i.e. “luxury travelers”, “green-living enthusiasts”).
  • Life Events – This audience type targets users who are in the midst of or have an upcoming major life event. We use “Life Event” audiences to target users who are “Moving Soon”.

The Results

In order to evaluate the effectiveness of these different audience types we first examined click-through-rate, cost-per-click, and cost-per-lead. That data is presented below. One thing to keep in mind – in the context of Gmail campaigns, a click refers to an email open, not necessarily a website session.

Interestingly, Affinity Audiences appear to do quite well in this accounting, with the lowest CPL by a wide-margin, as well as an exceptionally high click-through-rate. This is somewhat counter-intuitive, as Affinity audiences target proxy audiences – those with habits and behaviors we would expect to place them in our client’s targeted demographic, but without necessarily being actively searching for what our client offers. Their performance also deserves the caveat that we’re looking at a relatively low sample-size in this case. Consider, for example, that if there had been two fewer conversions, the Affinity Audience’s CPL would have been higher than the Keywords audience type.

Less surprising is the relatively high CPL of the Life Event audience type. This targeting method promises to identify users who are moving soon, but does not identify users who would necessarily be a match for the type of real estate offered by our client. This analysis suggests that in cases where advertisers are limited by budget, they might be best served by using the Life Events audience type to further narrow existing audiences instead of targeting a Life Event audience in and of itself.

The Keyword audience type distinguishes itself as one that was able to reach a sizable audience while maintaining relatively strong conversion efficiency. This suggests that advertisers should strongly consider keyword audiences when launching Gmail campaigns, and perhaps even expanding keyword audiences with a broader keyword set than they would normally consider.

Let’s bring in the session-quality data and see how it compares:

This data runs counter to much of the conversion data that we viewed above. Here, Affinity audiences seem to produce relatively poor sessions, while Life Event and Remarketing audiences produce relatively strong sessions. The Affinity and Remarketing data is a little easier to explain given the relatively small sample size – it’s completely possible that over 250 or 350 sessions we could see anomalous conversion rates that aren’t necessarily predictive of future success. In the case of the Life Event audience, however, it’s a bit strange that we’re seeing low bounce rates and users spend a good amount of time exploring the site without ultimately converting. One potential explanation is that targeting those that are moving soon does successfully procure traffic of users interested in real estate, but upon visiting the client’s site and learning more the users find that they aren’t a good match for the type of real estate our client offers. That hypothesis would be consistent with both the poor conversion numbers as well as the strong session-quality metrics.

Conclusions

While advertisers will be best-served by considering their own client’s particular context and running experiments to determine what audience targeting methods work best for them, I hope that the data presented above is useful for designing audiences in cases where budgets and time are limited. Our data suggests that keyword targeting can be an effective targeting method for Gmail campaigns that need to generate a good deal of volume. It also suggests that advertisers who are ultimately concerned with conversions might de-prioritize Life Event targeting, with the caveat that targeting by Life Event may very well be an effective method to further narrow existing audiences.

Questions? Comments? Please reach out on Twitter to @ppchero!