Understanding and quantifying the benefits of advertising is a problem as old as advertising itself. The problem stems from the many purposes advertising serves: building awareness of products, creating brand equity and generating sales. Each of these objectives is not easily measured or related to the advertising that may have affected it.
First, there has been an explosion in media alternatives from the traditional standbys of television, radio and print into a broader spectrum of both offline and online options, with the Internet clearly being the most visible example of this change. The choices within each medium have also expanded in an attempt to reach more targeted audiences. Television, for example, has burgeoned from three primary networks to literally dozens of mainstream cable channels, all capable of reaching large audiences with brand messages or product promotions.
Hundreds of new magazines now serve many special interest groups, while Web advertising presses the edge of one-to-one marketing. In addition, more and more companies are using integrated, multi-media strategies to reach their desired audiences, layering broadcast advertising over direct response campaigns or combining online with offline campaigns. All of this is making it harder to separate out the individual influences of each advertising effort.
Second, most companies are no longer satisfied relying solely on traditional methods of measuring advertising’s effectiveness, namely awareness surveys and tracking studies. They now want more precise and concrete evidence that their marketing investments are paying off. It is one thing to know how memorable an ad might be or how potential customers feel about a company or its products, but it is quite another to quantify the sales and profitability impact that advertising might produce. The difficulty is that measuring these effects may involve tracing advertising's stimulus through a behavioral chain of events that eventually may culminate in a sale long after the advertising has been delivered.
Third, companies today are in a more competitive and faster-paced environment than ever before, accelerating the need to understand the consequences of their marketing efforts. Marketers simply do not have the luxury any more to rest on their laurels or to assess how a set of campaigns performed months after they have concluded. The marketplace is moving so rapidly in many cases that knowing what's working and what's not almost as fast as it is happening has great value.
For these reasons, marketers have begun to use new measurement tools to help them estimate the impacts of their advertising efforts. One such technique that is gaining in popularity is marketing mix modeling, which when used prudently can add new insight into how to maximize business results from a range of marketing investments.
An Approach to Evaluating Advertising Effectiveness
While advertising may have several objectives, ultimately marketing and business executives want to know, “How has advertising contributed to sales and ultimately to the company’s bottom line?” Since we can’t ask consumers tell us what made them purchase a product, we have no choice but to make this assessment by other means.
The first thing to recognize is that advertising is only one of many marketing elements that affect sales. Other elements include pricing, promotional offers, product attributes, and reactions by competitors. In addition, external factors, such as macroeconomic trends and seasonality, are likely to affect response in the market. Consequently, the response to advertising can best be evaluated by techniques that allow researchers to account for the mix of marketing activities and other external factors.
Econometric modeling techniques, especially time series regression, provide a well-structured means of evaluating the impact of advertising on sales by isolating key explanatory variables and holding constant certain variables that may mask the effects of advertising. This is the technique used in marketing mix modeling, which has long been applied in the consumer package goods industry where the direct relationship of advertising to sales is weakest.
Under the right circumstances, we think marketing mix modeling is a useful way to assess the effectiveness of advertising, particularly when a company uses an integrated multimedia strategy. But we also recommend that marketing decision makers use additional information in deciding how much advertising to do and what the mix should be. In fact, the ideal way to manage marketing performance probably involves a combination of statistically robust modeling, traditional advertising tracking studies and other marketing analyses. Essentially, we believe that marketers are better served by using as many tools as they can in tackling this complex issue.
Advertising Carryover and Wear-Out
Another important consideration in quantifying the effect of advertising is measuring its carry over effects or persistence. Advertising media deliver impressions at a point in time, but psychological measurements of stimulus retention show that advertising messages may linger in the consumer’s mind over time. Thus, this relationship must be understood and accounted for.
What is clear from models we have done is that the effect of advertising persistence differs by media type. For example, the impact of broadcast TV may persist for months, while the impact of online advertising is likely to fade very quickly. For other advertising media, maximum durations are typically measured in weeks.
Thus, each type of advertising investment is likely to have different persistence, and models of advertising impacts need to recognize and specify these differences.
Continuous Measurement and Forecasting
While marketing mix modeling has its pluses and minuses, we think there are specific uses that are highly beneficial for marketing performance management, such as when it is used to understand advertising’s effects on different consumer segments. Moreover, we think that, in order for it to have an substantive impact on a company's business, marketing mix modeling needs to be applied in a systematic and iterative process, where modeling accuracy improves and the insights become more granular as additional spending and response data are collected.
Marketing mix models also may be used beneficially to project the likely results of future marketing activities. This capability is a key component of marketing performance management because it helps the marketer understand what is likely to result from the money he is spending now or is about to spend – which is far more relevant than simply looking at past spending.
The models enable the marketer to forecast results across all media, as well as simulate outcomes based on a set of alternative advertising plans. An example of this is using marketing mix modeling to predict the volume of leads coming into a call center to assist staff planning, or likewise to predict the number of visitors to a Web site requesting information that will be fulfilled via direct mail.
Managing Marketing Performance
Historically, maximizing marketing performance has largely been an exercise in uncertain intuition and “gut feel” for marketers. Using marketing mix modeling can help marketers immensely, especially at a time when they are under mounting pressure to account for their actions and spending.
The benefits of assessing advertising effectiveness using marketing mix modeling will, no doubt, vary from business to business, depending on company size, the role of advertising, and other marketplace factors. However, our experience shows that this type of analysis can greatly enhance a marketer’s ability to evaluate how well his investments are doing, enabling him to adjust the amounts invested in advertising or their mix allocation. For some, this will translate into multimillion-dollar benefits, which will certainly get the CMO’s attention, not to mention the CFO’s.
