Traditionally, e-mail program measurement focuses on overt consumer responses: We know that a customer has opened, clicked, perhaps even transacted — but unless a coupon redemption is involved, our knowledge of why a response has occurred is often limited. With the focus in measuring e-mail programs shifting to addressing questions regarding the long-term value of a customer, more information is needed regarding how e-mail contributes to sustaining that value.
Having proven that e-mail works, we now must provide in-depth analyses into exactly what works so that marketers have the ability to design programs that continuously engage customers through better, more integrated communications.
To accurately determine the extent to which our communications are influencing customer responses, we must know more about the practical relationship between content and responses. The effective integration of relevant content and response data means that the focus shifts from studying isolated, discreet responses to determining the relative contribution of many different marketing tactics. This is a daunting task, given that e-mail content falls into the realm of “unstructured” data, and that such data exists independently of business information systems that contain customer data, including responses.
How is this accomplished? Broadly speaking, there are three requisite steps. A critical first step is to capture and quantify the content information contained in e-mail. Secondly, this information must be related or mapped to the appropriate consumer response and other behavioral data. Finally, the content/response mappings need to be interpreted in the context of business and marketing objectives.
How do we quantify e-mail content? E-mail communications are a form of unstructured data, consisting of variable content designed for many different purposes. The underlying structure of e-mail, however, adheres to a standard that is very rich in information and can be leveraged for a broad array of valuable analyses.
Unstructured data is loosely defined as data that exists outside of relational databases and not neatly organized into rows and columns. Unstructured data typically takes the form of text, but also includes graphics, images, xml, flash animation and so on. Any analyst who has struggled with unstructured data is painfully aware of the effort, time, and money required to translate that data into a useable form. But as a consequence we deal with incomplete pictures of consumer behavior and the extent to which our marketing efforts influence targeted behaviors. In focusing on the response, we are forced to make too many assumptions regarding the cause and have a limited view of the customer.
The business intelligence that stands to be gained from solving the problem of unstructured data justifies efforts to treat it. By analyzing unstructured data, marketers will be able to learn how to improve content development and discern emerging trends and critical issues for specific customer groups.
Current tools for processing unstructured data have been generally inefficient for many reasons: They are not customized for a particular application; they are too broad-based or lack context; they are not integrated into traditional business information systems. To extract valuable information from unstructured data typically requires an inordinate amount of manual processing. Yet we can adapt and apply developing technologies to new areas and specific business questions to render timely and actionable customer intelligence.
The solution entails creating context-specific concepts and relationships, with the result of enhanced classifications or taxonomies to provide a rich context and framework for the analysis of consumer responses. Armed with the appropriate frameworks, emerging technologies for processing unstructured data can be customized for meaningful and time-saving applications.
Integrating Content with Response Data
E-mail communications can be defined and quantified along many dimensions, including content (tone, layout, length, brand, subject line, format), timing and frequency (date, day, time), offers (number and type), call-to-action (number, order, position and destination of URLs), and so on. Once the content of an e-mail has been defined along the relevant dimensions or taxonomies, and translated using tools that capture and quantify the data that exists unstructured form, the next step is to relate these dimensions at an individual customer level to responses.
After all, the advantage of wrestling with unstructured data to leverage the valuable information contained therein is still only half the picture. Only when combined with overt consumer responses can we determine whether our tactical manipulations (such as design and offers) are eliciting the desired consumer behaviors.
There are two general categories of responses relevant to e-mail — those representing the consumer’s engagement with e-mail (e.g., opens, clicks), and those reflecting the specific behaviors targeted by the communication (e.g., transactions, conversions). Each response measure reflects different aspects of the tactics incorporated into the content of the e-mail. Effectively linking input and response data requires a framework to ensure that — in the context of our marketing or business objectives — we are aligning the right content with the targeted response.
Our contact strategies are becoming increasingly complex, with each communication or series of communications defined by multiple objectives. We are becoming less direct response-oriented, embracing a broader spectrum of relevant metrics that contribute to customer lifetime value. A framework that maps objectives, as defined by the tactics employed within the context of each message, to the relevant customer response metric will allow the examination of where in the contact/response sequence things break down or new opportunities are presented.
Interpreting, Applying Results
However, creating more data does not automatically increase business intelligence. By being able to capture the unstructured data contained within the context of e-mails and convert it into quantifiable, measurable components, and by relating to response measures, the potential is there: The next step involves turning that potential into reality through the enhanced interpretation and practical application of results.
Testing to identify the most important tactics to employ or populations to target will help realize this goal. The basic purpose of testing is to investigate cause-and-effect relationships, but testing must be done quickly, efficiently and economically to be effective. The current complexity of communications programs, with multiple objectives and tactics, makes testing which examines individual tactics in sequential fashion unrealistic. If we have wider access to content and response data, we need to be able to process, analyze and leverage results in ongoing programs. There’s no point to having more data if we cannot process, analyze and apply that data in ways that contribute to customer intelligence and create real business opportunities.
The ability to translate e-mail content into quantifiable terms will greatly increase the number of tactics available for testing, and result in a large number of treatment combinations. To test the available combinations using traditional test designs and methodologies would prove costly, time-consuming, and difficult to interpret or apply in any meaningful way. Screening tests (e.g., fractional factorial designs), which make effective use of both data and experience, are available to identify and distinguish the most important factors from less important ones. These tests are typically followed by more in-depth and focused testing.
To optimize e-mail communications and contribute to the continuous improvement of programs, analysts must be able to provide insights that will enable a determination of whether observable differences in customer behavior are a direct result of specific e-mail designs and tactics. Armed with the knowledge of the relationship between content and response, marketers will be able to accomplish these specific objectives:
- Isolate the components of e-mail programs that have the greatest impact on e-mail engagement measures, store activity measures, and possibly online customer satisfaction scores.
- Use the information to guide the development and evolution of e-mail programs and offers that meet customer needs.
Katie Cole is vice president of analytics and research for Quris, a customer-centric e-mail solutions agency forFortune 1000 companies. Cole has over 20 years of in-depth knowledge and experiencein data mining, statistical modeling, primary research and softwaredevelopment in a variety of industries including telecommunications,broadband, Internet and financial services.