How much are my customers worth? Easy to answer now but how about in five years? This three-part blog series considers some challenges in effective Customer Lifetime Value (CLV) calculation and usage, providing considerations and guidelines for structuring model design, foundation and application within your business to drive value from it.

So firstly, how do we model CLV? We see clients across all industries struggling with this question. But why, when the basic calculation seems simple?

CLV = (revenue – costs) x length of customer relationship

Unfortunately, the simplicity of the calculation belies the complexity of each component. For example:

  • Revenue: can I identify my customers individually? Attribute individual spend? Historically? Can I extrapolate their spend pattern?
  • Costs: which costs should I include – variable costs and/or fixed costs? How accurately can I assign costs to individual customers? What costs feasibly reduce with volumes?
  • Length of relationship: can I predict individual customer churn propensity? Can I access the necessary data to predict this? What about the time value of money? Is historical extrapolation appropriate? What about approaching market changes and my business strategy?

Challenges vary widely by industry, company, business model, technology base, business structure and any other business facet. For example, subscription style industries (e.g.  telecomms) traditionally benefit from better customer data, more predictable revenues / costs and historical behaviour, due to their interaction style with their customers. However, the growth of ecommerce and loyalty programmes are allowing retail-style businesses to redress this imbalance, gaining transactional, behavioural and customer data unprecedented in its detail.

 Model accuracy will always be limited by your constraints and means, but application of some simple guidelines can help you approach these challenges in a structured manner that clarifies design assumptions and decisions to provide a stable and robust model:

  1. Use an appropriate measure: a matter of much debate, with contribution margin and net margin the most common measures. The key is to pick the measure appropriate to the objectives of the model. In general, functions that use CLV for individual customer decisions (i.e. target or drop this customer) favour contribution, whilst strategic decision making (e.g. business case development) within finance functions prefer the absorption costing nature of net margin.
  2. Be ruthless with revenue and cost definitions: the boundary between fixed and variable costs is grey at best. All fixed costs become variable in the long term and some "variable costs" are unlikely to be reduced with customer / consumption volumes (e.g. will reducing customers really translate to lower contact centre costs? How flexible is your workforce?). Projected savings are only achievable if revenues and costs are defined according to the model requirements.
  3. Allocate revenues and costs at maximum possible granularity: the model should use value data at the most granular level possible, as the aim is to differentiate between individual customer values. Any aggregation in the model reduces its ability to meet this objective.
  4.  Be clear of model objectives: the applications of CLVs within the business are wide but not uniform. Different functions have different requirements and multiple models may be required. Effective stakeholder engagement and requirements gathering in the design phase is essential to drive adoption and acceptance of the outputs.
  5. Create a stable and well governed model platform: multiple models will naturally diverge but the fundamental assumptions and calculations of the model must be consistent for reconciliation purposes and to ensure the business acts in a cohesive manner (See next instalment for details).
  6. Select a model that optimises accuracy given your means and objectives: all models face constraints that naturally limit their accuracy, be it the data, technology or analytics capability within your organisation. What is important is that your model has the fundamental accuracy that allows it to gain buy-in within the business for its adoption and application.
  7. Understand the validity of your model and use it accordingly: many of the organisations we work with on CLV have models already but lack of faith in its outputs. This leads to dispersed and inconsistent usage within silos across the business. Key to overcoming this is how the model is 'sold' to its stakeholders – no model provides a perfect prediction of the future. What is important is that the business understands the reliable aspects of your model and what it can and cannot be used for. We see many models that provide good directional insight but are mothballed because they cannot be reconciled to an established metric.

CLV models vary hugely in complexity, predictive capability and their ability to drive value for the business. These can be enabled or limited by many factors, some inherent to the business and some which can be overcome. This instalment has touched on some considerations in model build. The next instalments in this three-part series will look at key considerations in founding a stable modelling platform and effective application of CLVs within the business.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.