Following our previous posts on data departments, I’ll skip ahead and discuss something a bit more practical in this one. Data-driven customer or customer strategy is an umbrella title for dozens of different solutions and activities that I highly recommend organizations undertake. However, in this post, I’ll do my best to focus on the most essential ones.
What We’ll Discuss:
- The customer persona – the central building blocks, including deep knowledge of the customers and their behavioral patterns.
- The value the organization can deliver to its customers.
- Managing and predicting customer lifecycle events – identifying and managing opportunities and risks.
- Marketing automation – Managing the customer journey, and customer-based marketing including personalizing the value proposition/offering for the right customer at the right time.
- Digital customer management – what you should put an emphasis on when managing your customers in digital.
Customer Persona / Profile
The most essential building block when discussing customer management strategy is how well you know your customers. At this stage, we ask questions like what motivates your customer? What are they looking for in their interaction with your organization? What’s the best way to communicate with them? How much can this customer be worth? What’s their potential? What’s their projected lifetime as customers? When do they come to us? And so on.
The data department should create absolute clarity and produce a definite answer about all customer aspects, including these. The deeper the customer profile/persona you manage to create, the better you can leverage your marketing and customer service systems to maximize the return on your investment in the customer for prosperity. In some cases, the data department will gather the bits and bites of data that already exist in the organization – such as customer segmentation – and on top of that, the department would suggest innovative solutions that will crystalize the company’s knowledge of the customer into one clear bottom line.
In this context, here are a few foundational elements that are a must for any organization serving customers or selling to them:
LTV – Customer Lifetime Value
The LTV model’s purpose is to predict how much (monetarily) would the customer be worth to the company over their lifetime as a customer. Why is it so important to predict your customer LTV? Here are a few reasons:
- Value-oriented marketing – Maximizing customer value across different segments and business channels. In the same vein, making use of the marketing automation system to optimize decision-making related to customer-tailored offerings or benefits.
- Measuring the success of business/marketing activities in terms of customer value – e.g. did last month’s campaign grow my customers’ LTV?
- Strategy and planning – identifying areas where the organization is interested in growing its customer value and defining short-term and long-term action plans and roadmaps to achieve this.
- Customer service – offering personalized service to customers with higher value.
The advantage of the LTV model over other models (like RFM analysis- calculating customer quality) is that it predicts the future and enables the organization to be proactive.
In practice, most organizations prefer to predict the customer LTV for a defined time period.
The LTV model includes many sub-models predicting customer behavior. In order to reach a good prediction of a customer’s LTV and the potential of the organization’s engagement with this customer, we would also need a customer segmentation model, customer churn prediction model, RFM, and so on.
The LTV model will empower the data department to gain an understanding of which parameters influence the customer’s value and which of the organization’s customer-facing actions, in designing its products, in its digital experience, etc – has the most positive influence on customer value.
In order to implement this model, the data department is required to work with predictive analytics and machine learning tools, so the LTV model will be a constantly-learning model that responds to changes in the market and in the company, and would be able to adjust predictions to the changing scenarios.
Multidimensional Segmentation
Another building block, which is just as significant, when getting to know the customer is multi-D segmentation. LTV tells us what the customer will be worth throughout their lifetime but leaves out who they are and what are their preferences.
Multidimensional segmentation will help us gain a 360-degree view of the the customer across dimensions like consumer behavior profile, demographics, lifestyle, stage in life (age, marital status, parent or childless, etc), and what helps us reach them (personalized content, personalized benefits, product quality, discounts, digital user experience and so forth.)
Hence, multidimensional segmentation allows us to choose the right sales/retention/nurture methods, in the right context and in the right timing.
For example, let’s look at the fashion brand’s campaign. The customer is vegan (=lifestyle dimension), leads a healthy life, spends a lot of time online, and enjoys premium products. They prefer text messages as a communication method and are more sensitive to content (rather than discounts). In this case, we will target the customer with the relevant products at the relevant time and with offers that will maximize their (emotional) shopping style.
Often, the segmentation models come before the LTV because this information may also be helpful in predicting their value, but not necessarily.
LTV may deliver a quicker “win” than the full multidimensional segmentation, which might require more resources and will prove its full value in the longer term.
The multidimensional segmentation will enable the organization to build a more precise customer strategy for the different segments and create unique offers for its best segments.
Of course, such personalization will strengthen customer engagement and leverage customer value for the company in the long term.
Additional models that are necessary when getting to know your customers:
Abandonment prediction – Predicting customer abandonment in the short and long term or predicting their leaving/pausing activity in one particular area of service or product segment. This is a basic model which is also part of the LTV model. This model is the foundation for proactive and responsive behavior in the churn world. It shouldn’t only predict churn but also include the reasons for it.
Share of wallet – This model predicts what is the share of the customer’s activity with the company out of all similar activities they have with the company and its competitors. This model is both similar and different from LTV in that it helps the organization understand the customer’s potential. The main difference is that this model may not tell us how much of the potential we can reach, because some of the customer’s activities with competitors may not be subject to change, and the model may not be able to weigh their full activity with competitors. On the other side, LTV weighs your company’s behavior with the customer and its impact on the value that you would be able to derive. Either way, both models have significant – though different – roles.
Digital behavior – This model deals with the behavior of the organization’s digital assets and is different from other models. I’ll expand further below.
Of course, there are many more important and diverse models related to customer profiles, but the ones we covered are the most significant ones when you first dive into this topic. These models are based on prediction and on the ability to learn from the past behavior of customers about the potential future behavior of “similar” customers. Of course, it’s key that all these models are implemented using a learning model that can adjust with time. One of the main challenges when building these models is defining the parameters to include (feature engineering) and defining the best model for the specific desired prediction. Today, we have tools like Data Robot that can be helpful because they independently and automatically select the best prediction model for the specific need.
Value Propositions:
Now that the organization has developed the ability to create in-depth customer profiles, the next question is what are the value propositions it can offer its customers, and what will those offers generate?
As we’ve seen in the previous section, an organization that knows its customers well and that can predict their value, can also predict which offers would generate returns and for whom.
Many organizations have a hard time mapping their customer offers and value propositions. These can vary between product quality, service quality, discounts and deals, personalized deals, professional content, loyalty benefits, digital user experience, and so on.
The data department’s goal here is to gather all of the above and connect the dots between the most relevant value proposition for the most relevant customer profile. To do so, they will use learning statistical models that can dynamically adjust the right offer for the right customer.
The data department will also provide tools for planning future rewards and offers, based on the segments that the organization wants to develop and promote according to its strategic and tactical plans (for example, as a response to competitor’s activities).
Many organizations make the mistake of creating only one type of personalized offer (i.e., personalized discounts), while their customers may not be interested in this type of offer.
Managing Customer Lifecycle Events
One of the most essential topics in customer management is identifying and predicting various events throughout the customer lifecycle. An “event” can be something as simple and tactical as abandoning an e-commerce shopping cart, or on the more significant side of the range – entering a new product line, marital status change, or any other life event that can pose an opportunity (buying a house, for example, can be an opportunity for Finance, Retail and Telecom companies).
Once an organization can define, identify, and predict events like these, it can also choose to prevent or encourage some events. This is how you can prevent abandonment or motivate the customer to make a purchase based on a life event.
There are two main data-based activities related to customer lifecycle events:
- In-depth analysis (done by data scientists) to identify events that happen to the company’s customers across segments (similarly to multidimensional segmentation).
- Models predicting essential events as triggers to customer-facing activities.
Marketing Automation
Marketing automation is a methodology that many technological products try to accomplish. This methodology entails managing customers in an almost completely automated way, responding to customer lifecycle events in real-time, and managing the company’s various customer journeys. Marketing automation is starkly different in digital vs non-digital assets.
The marketing automation system combines all of the above into a well-oiled machine that allows the organization to maximize the customers’ value and behave in a personalized and precise way with each customer type.
Different organizations use marketing automation tools for different use cases. Some organizations may use them to integrate data from different systems, define segments and clusters, and manage customer journeys in practice. Other organizations mainly use these tools to manage customer journeys after the other data infrastructures above have already been developed separately.
Digital Customer Management
The digital age has taken customer-based marketing to never-before-seen dimensions and has enriched organizations’ ability to provide personalized, customized value to each of their customers. Virtually every business nowadays provides services digitally or is on its way there. Digital is an incredible platform for maximizing customer value.
Digital marketing can be split into two categories:
- Inbound marketing – marketing within the company’s digital assets, while the customer is there
- Outbound marketing – marketing outside of those assets, as a means to beginning the customer into them
The inbound marketing world is layered and complex and involves behavior within the company’s digital assets (such as the sequence of actions conducted before a purchase, identifying the customer’s intention, etc). There are specialized tools that can recognize the customer’s digital behavioral patterns, immediately suggest offers, and even personalize the user interface for each customer based on their preferences to maximize their value.
Of course, digital behavior also feeds into all the other models we discussed and includes specialized customer journeys in the marketing automation system.
One of the infrastructural challenges the data department would need to figure out is connecting the digital data system with the overall organizational data constellation. E.g., connecting Google Analytics to the organization’s Data Warehouse.
In conclusion,
The data department needs to lead the above activities while also exploring and producing insights that will help the executive leadership create the customer strategy. For example: Developing a specific segment, investing in a specific segment, launching a new business line, etc.
I hope this article provides you with a glimpse into the world of data-based customer management. It goes without saying that each organization should invest in mapping its current customer management status and creating a data-based customer management roadmap that will generate value in “lighter” areas and also long-term wins in other areas. Such a roadmap may include the ingredients mentioned above, for each segment or customer type.
Keep following for future articles about:
- How do I even begin? Creating a data department roadmap that will deliver quick wins while establishing strategic infrastructure to serve you long-term
- Different Approaches to data department Structures and responsibilities
- Data-driven customer strategy
- Data-based risk management
- Technologies that can be your data department’s secret sauce
- Data monetization – selling data services and products
