It appears that marketers are beginning to appreciate econometric modeling. Indeed, 70 percent of the Association of National Advertisers members now use econometric modeling to evaluate marketing ROI, according to a new study by the organization. Many of these companies have only added the approach to their research toolkit in the last two years. Most likely, these newbies are finding significant opportunities for marketing improvement by simply identifying what isn't working for them.
But some marketers - mostly in the consumer packaged-goods category - have years of experience with modeling and have long since harvested this low-hanging fruit. But while these companies benefit from ongoing ROI scorecards and the organizational discipline they inspire, they are increasingly unfulfilled by the short-term view of ROI, and the lack of insight into how their marketing is producing the measured effects.
These marketers want a more complete view of marketing performance that helps them plan for the short- and long-term, and provides insights to drive growth. They are pursuing new strategies that combine econometric techniques with brand-tracking data and panel research. While traditional modeling looks at what worked in the short-term, these expanded approaches address other issues, including how specific techniques worked, for what length of time, and with what audience.
With an understanding of which marketing tactics worked overall, marketers can ask which tactics performed for specific audience segments. For instance, do men respond differently than women? Do income, age, or ethnicity change the result? A segment view of performance can provide important indicators of opportunities for improvement.
To construct consumer segment models, marketers must have sales (or a sales data proxy) at that segment level. Most can find the necessary data. Many retailers use credit-card sales, while consumer packaged-goods companies can purchase panel data. Financial-service firms and travel providers have named customer data. When none of these options exist, custom tracking surveys can fill the gap, so almost any company can get to segment-level insights with a little extra effort.
Traditional mix models can also be expanded to explain how sales are generated. Are marketing tactics generating new trials? Are current users purchasing more or purchasing more often? Understanding these purchase dynamics can provide insight and identify opportunities to improve marketing spend, targeting, and messaging. And again, the required information is usually obtainable, be it shopper data, panel data or brand-tracking data.
Perhaps the most frustrating limitation of traditional mix models is that they provide only a short-term view of ROI, and do not quantify the value of advertising's long-term contribution to building brand equity.
Almost all marketers believe in the power of marketing to develop long-term brand equity, and want to account for it in their models and decisions. They can - if they have a strong, rolling brand tracker incorporated into their modeling.
Incorporating brand-tracking data into marketing-mix models shows that marketing's impact on long-term brand equity and sales varies greatly. In some cases, the long-term impact is many times the short- term ROI. In other cases, there is virtually no long-term contribution. Either way, marketers need to be armed with this information to have productive budget discussions with their finance partners.
Consumer segment, buying dynamics, and brand-equity models each extend conventional mix modeling insights beyond simple ROI scores. But when they are combined together in a single linked model set, they can reveal the underlying drivers of marketing performance so clearly that new opportunities for growth are much more apparent. For marketers who have relied on mix modeling to squeeze efficiency from their working dollars, this can be a welcome change of perspective. For marketers new to modeling, this more complete approach can provide more provocative insights, and inspire greater confidence to act on the results.