As the old marketing adage says, retaining your existing customers is much less expensive (5X less) than acquiring new customers. Consequently, it is important to focus your marketing attention and budget on developing strategies to systematically maintain and grow your share of wallet.
Step 1: Developing a rich set of customer attributes and behaviors is a key component in this process; the goal is to develop a 360 degree view. This knowledge provides the foundation for analytical routines and models, allowing for a better understanding of your customers. Customer attributes should be created by combining elements from as many customer touch points as possible, examples include the aggregation of order histories, contact history, website and search behaviors, demographic and other external market appends. Below are some general categories:
- Historical spend amounts and associated trends or velocities
- Recency of purchases
- Visit or purchase patterns; timing and frequency
- Depth and breadth of products in a purchase
- Response to promotions or discounts
- Historical campaign contacts and responsiveness, opens and clicks
- Channel preferences
- Digital data such as website visits and patterns, page views and click behavior, social media interactions, etc.
- Demographics (age, gender, household income, home value, etc…)
- Distance to store locations
- Competitive influences
- Seasonal influences
Step 2: Turn these attributes into marketing strategies that help us retain and grow our customer base. Specifically, leverage the newly defined attributes in Step 1 to generate more relevant communications and offers to our customers.
Analytics to enhance future campaign performance
If your objective is to maximize marketing campaign performance, one approach is to develop a two-stage model. The first stage involves the creation of logistic regression model to predict response to a marketing communication. The second stage is to create an ordinary least squares regression model to predict spend.
The combination of these two models can be used to identify the most likely responders, with the highest spend, resulting in increased value for subsequent campaigns. There are a few implementation alternatives to accomplish your goal. The first is to multiply the two scores together to create a joint score. This score can then be used to rank individuals based on estimated values, which takes into account probability of response. The second is to create a matrix using the two original scores. The Y axis might contain the probability score, divided into quintile or decile groups. The X axis would represent the predicted value divided into appropriate groups. We can then calculate the response and value for each cell in the matrix and identify which to target for the next campaign.
Analytics to guide customer interactions (Retention and Growth)
Here are a few tools that can be useful in developing an optimal strategy to retain and grow your customer base.
- Behavior pattern detection or Association Analysis; Based on transactional purchase or web behaviors, we can discover combinations of products that frequently sell together. Identifying these discrete buying patterns directs the development triggers or next best action contacts.
- Develop models to predict future value or spend (e.g., next 12 months)
- Determine current value (e.g., most recent 12 months of spend)
Understanding a customer’s current value in combination with their expected future value can guide our customer management approach. We can identify our ‘best’ customers those that represent a significant portion of our revenue and profit streams, and subsequently focus on the needs of these individuals. We can increase the number of contacts and tailor the offer to high value individuals. Alternatively, we may reduce the number of contacts and offers to certain low value individuals. Ideally, as part of this mix, is to include a forward looking view on your customer base by creating a separate treatment plan for customers with low current value but high potential value.
Within a next best action or trigger program, we utilize transactional or online patterns to both tailor the offer and timing to an individual. A customers value along with their predicted response to our messages can be used to inform, adjust or arbitrate actions within our trigger framework.
Finally, we encourage test designs that include control groups, to optimize any of these growth strategies.