Skip to Content

If an ad doesn't increase brand awareness, it's not doing its job. That's why David J. Reibstein and William S. Woodside, Professors of Marketing at The Wharton School, shared the the importance of using the correct methods for measuring ROI to help marketers track digital success. In addition, they take a critical look at the way these strategies are currently put together and emphasize the need for a shift from short-term results to long-term goals.

Everyone wants to measure marketing returns. How else would one justify proposed marketing spending? And ideally, measuring ROI should make it easy to assess which types of marketing expenditures are most productive. But there are different ways to measure ROI, each with varying degrees of precision and different advantages and tradeoffs. Below, we discuss some common ROI measurement methods, and what exactly they do and don't tell us about our marketing efforts.

Measuring ROI should be straightforward. You spend x dollars on a campaign or promotion, assess incremental sales, look at margins on those sales, then divide the incremental contribution by the cost of the program. However, there are complications. The two biggest are a) the coincidental factors that impact sales, and b) time.

Marketers often run test markets that control for covariants. Doing so in the physical world, however, can be both costly and time consuming, and to run such tests for every campaign would be prohibitively cumbersome. Doing "test and learn" experiments online is easier, cheaper, and faster. But not everything marketers would like to test can be tested online.

Measuring marketing's impact over time is more difficult. The impact of coupons or other time-limited promotions is readily measured. But with longer-term advertising campaigns designed to create awareness or shift perceptions, there can be a considerable lag between running advertisements and any eventual purchases. Even the coupon with a specified time limit may lead to a purchase, but what about the subsequent purchases that never would have happened if the redeemer never responded in the first place? A campaign's real impact may be cumulative, so any single ad's impact becomes hard to measure. Also, an ad may only increase probability of purchase, and probabilities are very hard to detect. And even if a single ad impacts immediate sales, what about recurring sales? All of a sudden that "straightforward" measure of ROI has become quite cloudy.

Instead, many marketers use surrogate measures of ROI. The most common is measuring program-to-program cost savings. If by using different media you get x exposures at a lower cost per thousand ("cpm"), it is easy to calculate the cost savings. This is why so many television commercials today are 15, instead of 30 or 60 seconds. Advertisers are measuring cost per exposure, and 15 second spots are always cheaper per exposure. NASCAR built its success, in part, on the number of television exposures a brand gets on a racecar and comparing the cost to the cost of more traditional media. The same is happening with new media today. How many exposures does a banner ad get on a given page? These views are being priced below other exposures in traditional media, and sold as a "better return" per dollar spent. Obviously, this doesn't actually tell you whether these exposures led to any incremental sales. It also ignores the "quality" of each exposure. A racecar whizzing by at 150 mph with a logo on it probably does not have the same impact as a well-produced television ad. And like many ROI surrogates, cpm is a tangible, easily quantifiable measurement, but what it is measuring is not directly ROI.

A campaign’s real impact may be cumulative, so any single ad’s impact becomes hard to measure.

A second common approach, particularly in consumer packaged goods, is the marketing mix model. Here, one builds regression models with sales being the dependent variable, and marketing efforts the independent variables. One starts with a baseline - existing sales momentum - and then measures the "lift" created by incremental marketing efforts. The lift - everything above the baseline - is attributed to marketing efforts, as shown in Figure 1.

There's still no picture of the long-term impacts of marketing spending. What marketing-mix models miss is how much of the baseline may come from previous marketing efforts. Measuring lift alone ignores long-term impacts. This leads to over-investing in short-term marketing efforts and under-spending on programs with longer-term impacts. And indeed, the last two decades have seen a dramatic shift to spending on short-term promotions, with matching reductions in spending on longer horizon advertising efforts. Marketing mix models also miss competitive dynamics. As one increases spending in one area based on a marketing mix model, competitors are likely to increase their spending in that area too. In this scenario, any "lift" may quickly vanish.

Today's new media offers new opportunities for measuring return. It is now possible to measure actual conversion or sales, not just exposures. This makes calculating incremental contributions easier. But nothing comes without its downside, and even accurately capturing incremental sales still discounts critical long-term impacts of brand-building and awareness efforts.

Rather than hunting for better measures of short-term ROI, I'm coming to believe marketers should focus more on marketing's impact on more intangible longer-term assets like brand, customer equity, distribution, and employee recruiting and retention. These too are measurable, yet better reflect the value of both short-term and longer-term marketing efforts.

What's your favorite flavor of ROI?