Traditional brand owners and retailers are increasingly encroaching into the e-commerce channel—and for good reason. After all, digital engagement with customers provides companies with valuable data on consumer behavior that allows them to optimize marketing and product development. In addition, by operating their own sales channel, providers retain control over user experience and brand image. Enter COVID-19, and suddenly the Internet is rapidly becoming the shopping channel of choice for more and more consumers, a trend that is likely to persist beyond the pandemic.

That said, the new e-commerce players also face a challenge: winning and retaining customers is an expensive affair. That is why it is crucial for success to invest primarily in those customers who are lucrative for the company in the long run. It is important to understand these customers intimately, to engage them with the right channels, and to tailor offers to their context and needs. This can only be achieved by drawing on customer-related metrics—of which customer lifetime value (CLV) is first among equals—and by interlinking them intelligently as the foundation for effective and efficient marketing.

Digitally aligned companies and start-ups have long been successfully applying and refining this approach (see sidebar, “‘CLV is our core steering metric,’ Four questions for Emmanuel Thomassin, Chief Financial Officer of Delivery Hero”). Many traditional manufacturers and retailers, on the other hand, still have some catching up to do. To make the most of the CLV approach and use it to manage their e-commerce business, they should adopt a long-term strategy and proceed systematically in three steps: collect data, determine true customer value, and target investments to the most valuable customers.

Collect data throughout the customer journey

To estimate the current and future value of customers and keeping privacy regulations in mind, companies need to collect relevant data points on as many customers and their behavior as possible over multiple years. This is because the corresponding analytical models are dependent on the availability of sufficient amounts of information to identify relevant patterns. The greater the volume of data available, the more meaningful and accurate the analyses. Three categories of data are required:

• Transaction data such as shopping timeline, product information, prices, method of payment, delivery, or returns are supplied by the e-commerce platform and the connected financial systems.
• Demographic data such as gender, age, occupation, and place of residence are condensed into customer profiles in order to better predict future shopping behavior and personalize marketing actions.
• Marketing data such as search behavior, response to campaigns, and external online data help to flesh out the respective customer profile and, in turn, deepen customer knowledge, including as regards preferences or purchasing behavior.

Despite ample data, it is often difficult to clearly identify customers throughout the entire customer journey. This is partly due to purchases made across different channels, for instance, in the company’s own online and offline stores or perhaps through third-party suppliers such as retail partners, which often do not require registration (with an e-mail address, etc.) for identification.

Successful providers solve this problem with an integrated customer database (customer data platform) that can recognize customers even when they do not sign in. For this purpose, profiles comprising as many attributes as possible are created for visitors to the various channels (based on browser data, among other things). Then, returning visitors (including to different channels) are identified by matching them against the full array of profiles compiled. Aside from linking different data sources and formats, the customer data platform also enables the integration of suitable external systems as well as customer segmentation according to behavior and demographic data. Key steps in this context include anchoring the system’s continuous improvement, but also data use by the organization’s departments from the outset.

Determining the true value of customers

What happens to the data collected? Here, in the second step, is where customer lifetime value (CLV) comes into play. This is because it can be used to measure a customer’s value, in the long term, over their entire time as a customer of the company. This value is compared with the customer acquisition/ retention costs (CAC), i.e. the marketing investments made or planned that are necessary to acquire and retain the customer. Finally, both indicators are linked to derive recommendations for action with regard to strategic and operational decisions (Exhibit 1).

A distinction is made between three levels of complexity when modeling CLV and CAC:

The descriptive model calculates CLV using historical consumer data and identifies behavioral patterns of customer groups mostly through simple manual analysis. This comparatively simple method yields rapid results, but they are merely hypotheses and therefore of limited value; they can only serve as an initial indicator of CLV for potential decisions.

The predictive model uses historical data patterns to determine future CLV. Consequently, the results are more accurate and meaningful as the customer’s individual profile is factored into the equation along with their remaining time as a customer. Backed by this knowledge, CLV managers can make more effective decisions. However, this model requires more comprehensive advanced analytics capabilities, such as customer identification across multiple channels. For a 360-degree view, it is worth having complete historical customer data as well as regular updates of sales and cost data.

The operative model goes one step further: it automatically predicts CLVs using machine learning and makes initial recommendations for decisions, amplifying the CLV effect. In addition, predictive accuracy and decision making improve with each update. For the operational teams, this means that rather than elaborating decision recommendations, their primary job is to review and continuously monitor them. Yet, creating such models is a much more complex endeavor that can take months, if not years.

For all three models, continuous updating data and calculations is indispensable. For example, CLV must be adjusted after each customer purchase, but the CAC value must also be increased if, for instance, a marketing campaign is launched for a specific customer group. This is essential so that the data and the associated analytics results can be used for future campaigns.