A lot of marketers are grappling with the challenge of media/channel attribution in a social network setting and addressing the related issue of allocating resource across different vehicles and customer/consumer touch points. A recent example illustrates the importance of these decisions for marketers’ budgets and effectiveness:
A relatively young and dynamic company uses search engine marketing, blogs, RSS news feeds, social networking, podcasting and user-created content to interact with customers. Without the data and time to allocate resources according to some sort of marketing mix model, or use a heuristic method such as apportioning budgets relative to each revenue contributions, the company decided to perform a survey-based segmentation specifically addressing media consumption patterns and preferences.
The study revealed that segments of customers that still prefer print-oriented content, as well as customers that prefer newer media, such as video, and still others who prefer social media and word of mouth. These segments then served as a foundation for this company's media resource allocation strategy.
This is a simple, intuitive concept -- and a valuable one to remember. As we move into a world where communications are two-way and have the power to engage customers, we should be allocating resources around our customers' consumption of marketing, rather than older methods designed to push content at the most efficient cost.
While the idea is sensible, from a technical perspective, using segmentation for marketing media allocation is not trivial, even when -- as in this case -- the segmentation is built around media preference.
When we segment customers by their media preference, we must make sure that we are identifying not just consumption preferences, but also the relative effects of channels to customers' behaviors. For example, the segmentation should contain both descriptive constructs, such as customers' media preferences, as well as normative and relationship constructs, such as the relative importance of different media on consumers' satisfaction, loyalty, and future repurchase intentions. Some channels may be highly engaging from an interactive perspective, but they fail to promote true engagement through loyalty and repeat purchasing. In such cases, we do not want to allocate solely based on preference.
What we need to do is combine the elegant simplicity of that company’s segmentation approach with the rigor of more traditional marketing mix analysis, ideally by integrating them into a single framework that determines resource allocation. Coca Cola has an analogous problem, in that it develops many marketing mix models for various product categories and marketing regions each year, but both the quantity and quality of the data can vary significantly from market to market. In some markets, there might be no pricing data collected, or the price has been flat historically and no variation exists in the data.
In these cases, the Coca Cola marketing team will use marketing research and segmentation data (a conjoint study on pricing, for example) and estimate the price elasticity using a Bayesian approach where the segmentation or marketing research results are used as priors of the models. In this way, the assumptions derived from segmentation can be tested against real-world outcomes derived from sales data.
Even companies with access to rich behavioral data that they use to develop their media allocation could gain by supplementing their models with marketing research and segmentation. Neither the segmentation nor the marketing mix model is perfect. Both make assumptions and contain varying degrees of uncertainty. Although integrating results from these two approaches can be challenging, the results can be significantly better than using a single approach. From a practical standpoint, if the most effective channel or marketing vehicle determined by your marketing mix model (having the highest marginal elasticity, for example) is the least preferred channel according to your marketing research, you would be wise to pause and think it over.
In a world where communications increasingly seek to engage, rather than to merely sell, incorporating customer preferences and consumption of marketing media into your media mix allocation adds an important new dimension. Blending that focus with a media mix methodology that also makes decisions based on outcomes provides a new strategic orientation, one that is at once customer-centric, yet designed to build value for your business.
Hongjie Wang is vice president of Customer Analytics at Fulcrum, a company that specializes in advanced analytics, technology and multichannel program solutions for marketing.