“Half the money I spend on advertising is wasted; the trouble is I don’t know which half.”
A performance marketer proudly and confidently hands the chief marketing officer (CMO) a campaign report indicating that digital retargeting efforts have generated a 75:1 ROI. The CMO asks, “Why should I believe these results when no other channels included in the marketing strategy drive more than a 5:1 ROI?” Perplexed by the lack of excitement and gratitude, the performance marketer inquires further with the CMO about why the results were not impressive. The CMO reveals the ROI is too good to be true and raises suspicion that a large portion of sales revenue attributed to the digital retargeting efforts would have happened regardless of the campaign running or not. So how could the marketer prove to the CMO that the campaign did in fact generate sales revenue that is attributed to the marketing efforts?
Incrementality measurement enables marketers to understand the true impact their marketing strategies have on a customer’s decision to purchase. For instance, the performance marketer in the above scenario could have set up an A/B lift test for the campaign and compared performance of the retargeting audience exposed to marketing to a subset of the retargeting audience that was purposefully not exposed. This would have enabled them to determine what share of revenue attributed to this already high-intent audience was driven as a result of media, and which share would have happened regardless. In doing so, it’s almost certain that the resulting ROI measure would be substantially lower than the initial read of 75:1, potentially even less efficient than the 5:1 standard across other channels. However, this would have provided a far more useful ROI metric to measure performance and convincingly prove to the CMO whether the marketing strategy is working.
With the breadth and depth of data now available in digital marketing, it can be tempting to invest heavily in niche audiences comprised of consumers with a strong natural propensity to purchase for a brand. However, without a proper measurement strategy, marketers leave themselves vulnerable to over-investing in tactics that have minimal or no influence in driving incremental business results. As the performance marketer experienced, heavy investment toward high-intent, hypertargeted audiences can often lead to illusory results that greatly overstate performance. This happens because a significant concentration of consumers within these audiences would perform the same action with or without ad exposure, allowing the ad to take credit for an outcome it played no role in driving. This is not to suggest that investing in granular, high-propensity audiences is a bad thing. In fact, it can be extremely valuable to reach these audiences. The key is to do so responsibly with the right measurement plan.
When evaluating and planning a digital media strategy, it’s imperative that marketers ask themselves, “How are we going to measure incremental ROI?” Is it by utilizing a test-and-learn approach geared toward driving a quantifiable business outcome? If so, great. If not, it’s time to develop one. Otherwise, marketers run the risk of driving suboptimal results at best, based more on correlation than causation. The good news is there’s no shortage of ways to achieve this in the digital marketplace. To be successful with any test, keep these important considerations in mind:
Identify the best lift measurement solution(s) for your digital campaign
Traditional approaches like the utilization of the public service ad (“PSA”), geographic splits or CRM user holdouts can be highly effective if executed with the right degree of rigor. Additionally, in the last four to five years, newer and more innovative approaches have arisen that mitigate some of the drawbacks of these traditional approaches. For instance, Facebook and some programmatic platforms have developed methodologies that can dynamically hold eligible users back for a control group without the need to allocate otherwise working media dollars toward a PSA or other type of placebo.
Measure what matters
Select a business outcome that you‘ll be able to measure success against for the test. Ideally, this is a direct business outcome. If not, try to identify a strong leading indicator or proxy. It‘s worth noting that offline data is becoming increasingly available for measurement in digital advertising platforms. Once you‘ve done this, identify a KPI goal against that outcome. Keep in mind that positive lift on its own is not always a successful outcome. The lift needs to translate to a goal focused on incremental return for your business, such as ROAS or CPA among others.
Effectively balance audience reach with significance
Carefully consider parameters, such as exposed vs. control breakout percentages, and test duration needed to attain statistically significant results. Typically, the higher an allocation can be against the group you mean to expose, the less opportunity you are leaving on the table. Therefore, rather than defaulting to a 50/50 exposed vs. control breakout, work with your data teams ahead of time to determine an ideal split that allows you to garner significance while also reaching a meaningful portion of your audience.
Generate actionable results
Proving campaign performance success is great, but success (or lack thereof) should also make you smarter for future activation. Consider developing additional test cells, such as creative or audience segment, to understand which pockets of media are more incremental than others. This will enable you to derive more actionable optimization opportunities as a result of your testing.
All incrementality measurement solutions have their pros and cons, and no one solution will solve for all digital channels. The key is to not let perfect be the enemy of the good but rather to define a business goal and utilize an approach that puts you in the best position to measure and optimize performance against that goal.