For this blog I would like to dive into the literature and briefly explore the concepts and strategies described in the “Nine Easy Plays to Get Started with Predictive Marketing Playbook”. For further reading I recommend the eye-opening book “Predictive Marketing, published on August 6, 2015” written and copyrighted by Ömer Artun and PhD. Domique Levin.
Even though the book was published 5 years ago, it has never been more relevant than now. I truly believe the book was ahead of its time and even today many companies have not yet evolved their marketing-game to the level as described and conceptualized by the authors of this book. Still today predictive marketing is hot topic among small and large companies. Early adopters such as Amazon and Netflix already reaped the benefits years ago. According to the book, reasons why you should care about Predictive Marketing are: (a) customers are demanding more meaningful relationships with brands, (b) early adopters show that predictive marketing delivers enormous value and (c) new technologies are available to make predictive marketing easy. Especially the latter is now truer than ever.
I am providing you with a sneak peek into the playbook and chunked it up into digestible bites. Further I want to investigate how SAP Marketing Cloud could help a modern market(e)er to get started with putting the playbook into practice. It would be too much for this blog to explore all nine plays in the playbook so instead I have picked four actionable plays.
OK, without further ado, let’s get to the four selected plays!
Play One: Optimize Your Marketing Spending Using Customer Data
In this play we explore how to allocate marketing budgets wisely. Many market(e)ers are allocating their budgets to the most important sources of revenue, the best performing channels, or the best performing ad-words. However, through predictive marketing, it would be interesting to allocate budget to the right people instead of to the right products or channels. This can be done by following this framework:
- Invest separately in acquisition, retention and reactivation.
- Differentiate spending on high, medium and low-value customers.
- Identify the products that bring the highest lifetime value customers.
- Identify the channels that bring the highest lifetime value customers.
As we know, it is cheaper to retain your customers than to acquire new ones or reactivate dormant customers. This has mainly to do with the number of marketing touches that are involved to acquire a new customer. Therefore, it is very critical to focus on customer retention and allocate your budget accordingly. Marketing Cloud capabilities such as Dynamic Customer Profiling in conjunction with Segmentation enables the market(e)er to identify those specific customer groups. This provides valuable input for the Marketing Planning and Performance processes. Eventually, with the budget allocated to the right marketing activities the campaigns on your marketing calendar will target the right people with the right priority. For example, turning one-time buyers into repeat buyers and reactivating dormant customers to buy again.
For optimizing the acquisition budget, it could be wise to make decisions based on payback time or customer value. The spend management capabilities of Marketing Cloud integrated with SAP ERP helps to determine the acquisition costs based on the actual spend. Based on the acquisition costs and the actual revenue, the payback time and ROI can be determined. This provides powerful performance insights on which acquisition activities and sources have led to the greatest customer value at the end of the evaluation period. Going forward, when allocating the customer retention budget, it is wise to think about different value-based segments. It goes without saying that it is more costly to lose a high-value customer than losing a low-value one.
Play Two: Predict Customer Personas and Make Marketing Relevant Again
The next step in optimizing your segmentation models is the application of “clustering” based on predictive models powered by machine learning. This is a very powerful tool that allows the market(e)er to identify homogenous groups within their customer base. Clustering is a tool to develop strategies based on a clustering schema to specifically target a group of new or existing customers matching a certain persona. The book describes three types of personas, namely product-based, brand-based and behavior-based clusters. A product-based cluster is a group of customers related to the (type of) products they tend to prefer or combination of products they tend to buy together. This schema is especially useful for retailers who sell products in many different categories and want to identify crossover shoppers between categories. Brand-based clusters identify the brands groups of people most likely will buy or brands they are not interested in. Many retailers have discovered that in many cases customers have more affinity with a brand than with a product type. Lastly, behavior-based clusters look at the behavior customers are showing while purchasing a product. For instance: which channel are they using, how frequently are they buying, are they buying only discounted items (discount junkies), are they returning products frequently (returnaholics)? This type of predictive analytics is the cornerstone of the buying propensity models SAP Marketing Cloud. Together with the segment builder it becomes a very powerful tool. With this we can create multi-dimensional segments that can precisely target the right people with the highest propensity scores thus enables the market(e)er to effectively engage customers with a relevant message.
Play Five: Predict Likelihood to Buy or Engage to Rank Customers
Buying propensity, or likelihood-to-buy models sound like magic. It is almost like looking through a crystal ball and predict that a customer will buy a product or service based on the customer behavior. This is very valuable information when optimizing email frequency in campaigns and send special offers depending on the likelihood to buy. If a customer is already motivated to buy, then he/she do not need much extra incentives to pull the trigger. Whereas customers with lower buying propensity score need a better offer to gain interest on buying something they did not want to buy before. This targeted approach optimizes both margins and revenue, as well as customer retention. This is all possible by comparing prepurchase behavior of prospects with prepurchase behavior attributes of many customers who bought the product. It is all about big data and machine learning to determine an accurate buying propensity score. Besides predicting the likelihood to buy, it is also valuable to predict the likelihood to engage. With these models we can predict if a customer will open the e-mail and potentially click through, or if the customer will likely unsubscribe and opt out for further communication. By intelligently target your campaign to recipients with a high propensity score we can lower unsubscribe rates and increase the chance of a successful outcome. An opt-out is more costly down the line than you might expect as it greatly reduces the customer’s future lifetime value. With the Predictive Studio and Score Builder in SAP Marketing Cloud we can build these “Likelihood” models and apply predictive algorithms and machine learning to calculate propensity scores. These insights can be used in reports as decision support but are also tightly integrated with applications throughout the solution. Think about customer segmentation, campaign execution and lead management where propensity scores are most valuable and actionable.
Play Six: Predict Individual Recommendations for Each Customer
Systems to make personalized recommendations have been around for about 20 years. That being said, it has not been widely adopted and implemented by many companies. There is so much potential left, especially when combined with predictive analytics. Making personalized recommendations can be divided into three parts: sending the right content, at the right time, within the right context. Today’s recommender systems can make use of big data and predictive models to see patterns in behavioral data of other people to predict an accurate individual recommendation. With this data insight we can either attempt to up-sell or cross-sell with relevant recommendations during a purchase or next sell after a purchase have been made. By integrating SAP Commerce Cloud with Marketing Cloud we can leverage the standard and custom recommendation scenarios within the commerce platform. And it works both ways, as Commerce Cloud is a valuable data source for Marketing Cloud when it comes to profile enrichment and provide tons of transactional and behavioral data. If the content you are presenting is relevant and correctly timed, the content design becomes less important.
This sneak peek into predictive marketing provided concepts and ideas on how to intelligently plan your marketing spendings, how to make your marketing message relevant again, and address it at the right time, to the right people with the highest likelihood to buy or engage. Predictive marketing becomes especially powerful when combined with multiple applications within the CX-landscape. This could dramatically change the way you do business today! If you are already excited and want to get started, contact us.