As digital-data tracking and availability has evolved, marketing-mix modeling techniques and depth of insights have also evolved, allowing marketers knowledge of consumers they only once dreamed of.
Pixel and cookie tracking by ad servers (such as DoubleClick or Atlas) reveal a vast amount of details about online consumers. Pixel tracking provides information such as IP address (which reveals geographic locations of users), as well as time and date stamps. Cookies tell us what consumers clicked on, how much time they spent on the site and more.
But looking at these as isolated data points, and without linking them to incremental outcomes, only scratches the surface of what can be accomplished. Integrating this digital data into marketing-mix models provides the complete picture of the marketing landscape and the relative performance of marketing vehicles and tactics.
More data from more sources, like Omniture, DoubleClick DART, Google Analytics, ComScore, Facebook and Twitter, enables more channels to be incorporated within marketing-mix models with more granular detail. Marketing-mix models are ideal for analyzing these disparate sources of data because evaluating such data streams in isolation risks errors such as last-click attribution errors.
Case Study: Choice Hotels
Choice Hotels International, one of our clients, provides an example of how marketing-mix models can help.
The company has a history going back to 1939—you can't get much more bricks-and-mortar than a hotel chain. However, Choice Hotels recognized early on in the digital revolution that Choice's digital presence was becoming almost as important as its real-estate presence. Anyone who has used a site like TripAdvisor to research hotels, or Expedia or Kayak to research rates and book rooms, knows the importance of digital to the hospitality industry. As such, Choice Hotels endeavored to leverage marketing-mix models that integrate digital data to give a full picture of the marketing landscape.
Consumers are spending more time online and with mobile, and therefore, individuals are being targeted across multiple online channels in addition to traditional media channels. Without the implementation of a program that measures digital at a granular level, Choice Hotels would only be able to analyze campaign performance within simplified, distinct silos. Therefore, Choice Hotels sought to implement marketing-mix models that incorporate both online and offline attribution.
Last-View Attribution Errors
One particular problem of traditional digital analytics is last-view or last-click attribution error. This happens when a consumer action or purchase is attributed only to the last ad impression he saw, rather than others he was exposed to in multiple channels before that. Consider the case of a hotel chain. Even if the consumer discovered the hotel on TripAdvisor and came across its social-media efforts or search, if the last thing she clicked on before purchasing was a banner ad, last-click attribution would assume all the credit for the sale goes to the display ad.
Implementing complex digital data in marketing-mix models allows a host of questions to be raised—and answered. What is the optimal allocation between offline and online marketing efforts? What is the optimal allocation for online media (search, display, email and social media)? What are the more granular aspects of digital media that can be optimized? (Keywords, creative, ad size and placement, day of the week, mobile, social?) What are the interactions between our offline and online spending? Is offline media improving our online efforts, and by how much?
Answering those business questions allows a company to fine-tune its media mix to enhance overall marketing performance. And in Choice Hotels' case, with last-click attribution errors out of the way, it was able to make better strategic and tactical decisions to improve marketing effectiveness and sales performance.
The internet has fundamentally changed the way consumers learn about, interact with and shop for the brands that they enjoy. Savvy marketers have incorporated the successive waves of digital data into traditional marketing-mix modeling. As we move forward, more complex marketing-mix models will reflect the reality of that changing marketing landscape, and allow marketers to take advantage of the results that come from effectively coordinating multiple media channels.