Improving Engagement and Performance in Digital Advertising

Improving Engagement and Performance in Digital Advertising

          
Title image

Improving Engagement and Performance in Digital Advertising

  • Add To Interests
  • SAVE CONTENT
  • PRINT
  • PDF

  • Related Articles
    How to Improve Engagement

    Advertisers and their agencies have the tools they need. By adding advanced techniques to their campaign mix, and reallocating budgets accordingly, the advertisers in our study were able to produce better results in numerous areas—including both reduced costs per action and increased action rates—while maintaining the scale of their campaigns. (See Exhibit 2.) A major automaker, for example, reduced its CPA by 32 percent, its cost per view-through by 29 percent, and its cost per click by 6 percent—while increasing its click-through rate by 36 percent and its view-through rate by 81 percent. (See the sidebar “Our Methodology.”) Even those advertisers that thought they were operating at a best-practice level or using the latest techniques were surprised to see such gains.

    exhibit

    Our Methodology

    We used an A/B testing methodology to analyze the impact of advanced techniques on the effectiveness of digital-marketing campaigns. Our study involved five advertisers in four different industries in Europe and North America. For each advertiser, we ran two campaign scenarios—a control and a test group—in parallel for four to six weeks with identical budgets that ranged from $30,000 to $80,000. Both campaigns were expertly planned and executed, and exclusion rules ensured no cross-contamination of users between the groups.

    Each control group was designed to be representative of a typical display-advertising campaign and was based on past campaigns that each advertiser had run. The control groups’ campaigns used primarily site-based and standard behavioral targeting techniques. Our test scenario employed the same techniques as the control group but added advanced targeting techniques, such as display remarketing from search ads, video remarketing, and behavioral analytics. In the test groups, advanced techniques accounted for approximately one-third of total spending.

    The control and test groups were treated equally at each step of the campaign. For the first week, all techniques were allocated equal budgets. By the second campaign week, the parameters related to individual techniques—such as bid levels and budget pacing—were adjusted to maximize performance. For the final weeks of the study, the worst-performing techniques were dropped and their budgets reassigned to the stronger performers.

    Our analysis of performance included clicks, view-throughs, conversions (as defined by the advertiser), and, when possible, user-engagement metrics on the advertiser’s website, including time spent on page, number of pages visited, and bounce rate. The control and test groups were filtered for statistical biases unrelated to advanced targeting. The statistical significances of the performance uplifts among the control and test groups were tested using a resampling approach—and the observed differences in cost per action, cost per click, and cost per view were each found to be statistically significant at 99.9 percent using a one-tailed test.

    For a subset of the study, a third baseline scenario was run in parallel to the test and control groups. This scenario included targeting techniques from the control and test groups but displayed a charity ad that was unrelated to the campaign, rather than the campaign creative. This baseline scenario provided a measure of how many of the users being targeted in the control and test scenarios would have viewed through, engaged, or converted even in the absence of a relevant ad.

    The “freshness” and “completeness” of data available for campaign targeting and optimization is critical to the performance we achieved. Advertisers with a data strategy that ensures data is available in real time (for example, right after a consumer has clicked on a search ad)—and not lost across channels—will have a significant advantage over those that do not. They can target consumers at specific stages of the purchasing journey in ways that those using standard techniques cannot—leading to a higher level of conversions. (See Exhibit 3.)

    exhibit

    Advanced targeting also enables the identification of users who are more readily and deeply engaged. In the five-week period of our study, a large North American bank used video and display remarketing from search ads, as well as behavioral analytics, to reach more engaged users who subsequently spent 30,000 more hours and visited 1,000,000 more pages on its website, compared with a control group not targeted using advanced techniques. This means that the average targeted user who then came to the bank’s website spent 30 extra minutes browsing, visited 14 more pages, made seven additional site visits, and had a 10 percent lower bounce rate.

    Getting the most out of advanced targeting requires a thoughtful approach to data—in particular, choosing the right metrics for optimization. For example, our tests indicate that view-throughs often have a higher correlation with conversions than clicks—and view-throughs provide more data from which to optimize. But many advertisers have no way of tracking this critical metric and tweaking the campaign based on its performance. One information-services advertiser achieved a reduction of approximately 50 percent in its cost per conversion compared with its best-ever historical performance by employing advanced targeting techniques and by switching optimization to focus on view-throughs. Clicks were not strongly correlated to conversions for this advertiser—and compared with the alternative of using conversions as the metric for optimization, view-throughs provided many more data points during the campaign, thereby enhancing the advertiser’s ability to separate real performance boosts from digital noise.

    Sophisticated advertisers—both brand and performance focused—have created “proxy” metrics based on customer engagement and enabled by programmatic buying when the metrics linked directly to campaign goals (such as sales) may not be available or significant. For example, some advanced brand marketers that do not have their own direct e-commerce presence optimize digital campaigns on a cost-per-engagement basis in which prolonged interactions with key Web content are deemed to better correlate with campaign goals than simple clicks or views.