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
  • PDF

  • Related Articles
    Five Steps to Improved Performance

    Digital advertising is a complex and often inefficient business, but powerful solutions exist for those companies that make the effort to apply them. Last year, BCG mapped and measured the end-to-end processes of 24 digital campaigns across
    15 European advertising companies. We found that complexity too often reigned. We identified more than 25 common pain points across the value chain caused by various stakeholders, processes, and tools. Those companies that had determined to get their arms around the complexity, however, realized a saving of staff time of up to 33 percent in their campaign operations. (See Efficiency and Effectiveness in Digital Advertising: Cutting Complexity, Adding Value, BCG Focus, May 2013.) Our 2014 study demonstrates that similar substantial uplifts are possible in targeting and engagement. To achieve them, advertisers need to do five things.

    Adopt a unified technology platform. This is a first step that helps pave the way for other improvements, including the use of advanced targeting techniques. Unified technology platforms provide a single user interface and make it possible to source data from a single pool, eliminating the need to reconcile, consolidate, and transfer data among multiple sources. This provides the data “freshness” and “completeness” crucial to achieving good performance in targeting. Unified platforms also allow remarketing lists to be used across tools and channels in advanced targeting techniques such as display remarketing from search ads. An advertiser in our study used a unified platform to make fresh search data available across two separate teams developing search and display campaigns. This allowed for immediate retargeting of consumers who had clicked on a search ad for the brand. Using one set of tools also ensures that teams aren’t cannibalizing performance by bidding against themselves—a phenomenon we witnessed several times during the study.

    Among other benefits, unified platforms allow campaigns to launch faster. Agencies can also spend more time gaining insights and fine-tuning campaigns, rather than wasting time on manual downloads and data consolidation.

    Implement advanced techniques. Three of the four advanced targeting and engagement techniques tested in our study resulted in substantial improvements in almost all key metrics for all advertisers. With display remarketing from search ads—serving display ads to users who have clicked on a paid search link purchased by the advertiser—marketers can target an additional group of consumers actively searching for a product or service. This technique drove 24 percent of conversions in one test campaign, with a CPA that was approximately half that of the best previously achieved by the advertiser.

    Video remarketing involves serving ads to users who have interacted with the advertiser’s video channel. (Interacting, in this case, means either watching a video or subscribing to a channel.) This drove the lowest cost per click for two of the adver-tisers in our study. Moreover, this technique reaches users early in the purchasing journey, when they are only just starting to engage with the brand—often acting as an early driver of conversions overlooked in common last-click attribution models.

    The use of behavioral analytics reaches users that have not only visited the advertiser’s website but demonstrated particular behaviors while there, such as spending a certain amount of time or visiting a certain number of pages. When one advertiser remarketed to users who had visited two or more pages, CPA was 26 percent lower than simply remarketing to all website visitors.

    Among the advanced techniques we tested, only targeting third-party look-alikes—which relies on identifying third-party lists of users with characteristics akin to an advertiser’s own first-party targets—failed to generate significant improvement. This may be because the quality of third-party lists can vary considerably and our test period did not allow for the extensive trial and error of various lists.

    Attack fragmentation. Advertisers generally look to agencies to tackle the fragmentation problem, but it is the advertisers themselves that often mandate multiple strategies and teams to begin with. Advertisers will likely have to take the lead in addressing the issue, which will be a tough challenge. Many bad practices are entrenched, and campaigns rely on multiple parties in their development—some of which have little incentive to change behavior.

    Still, there are some basic, no-regrets steps any company can take that will help. These include the following:

    Root out the causes of fragmentation. Select a typical digital campaign and map out all the various internal and external people, teams, agencies, and tools involved. This will show where companies sit on the “fragmentation meter” and let ad- vertisers identify practices that can be combined or consolidated. (See Exhibit 4.)


    Break down silos. Advertisers and agencies need to make sure everyone involved in running campaigns communicates with each other regularly to discuss how the strategy gets translated into execution.

    Connect the data strategy with the overall strategy. Advertisers need to connect their data strategy with their overall strategy, putting in place tools to collect, analyze, and leverage data in a centralized manner. Advertisers also have to make sure that they have the metrics that correlate to and drive the key strategic objectives of the campaign. Having one centralized source of data linked to other tools ensures it is available when and where it is needed.

    Get closer to your data. This is one of the marketer’s most potentially valuable assets. Advertisers should ensure that no major digital assets are being underleveraged (video, search, and website)—and that no data is lost along the way. Marketers can test the statistical significance of their metrics and ensure that the important progress indicators of the campaign aren’t just noise—and if they are, test whether better proxy metrics can be created.

    Use multiple attribution models. For example, our study found that video is often undervalued in a “last click” attribution model, despite playing a critical role early in the purchasing journey. Further, broad site-based targeting tended to get too much credit.

    Bring the math men and women to the table. Marketers and agencies need to augment their digital skills. Strategy and creativity will always be critical to successful campaigns, but the ability to effectively collect, analyze, and use data today is no less essential. Too often, however, companies regard this capability as a stand-alone function segregated from mainstream campaign development. Mathematicians, statisticians, and other analytics experts need to be part of the campaign team and part of ongoing campaign improvements and evaluations. Advertisers should look to structure and train their internal teams to leverage these capabilities—and to work with their agencies to fill skill gaps.

    Test and learn. The Internet is the best marketing laboratory yet invented. Direct marketers know this. Because their ultimate metric is online sales, they are adept at testing one technique against another and tweaking campaigns until they get the results they want. Brand marketers can take a page from this book, substituting their own metrics—such as average session duration as well as graphic interaction and information downloading—for direct sales. Our study tested individual campaigns one against another with the goal of seeing whether using advanced techniques could improve effectiveness. We came away convinced that by experimenting further over time, we could readily deliver continuous incremental improvements on top of the already substantial increase of 30 to 50 percent that we achieved.