← All projects

Case Study

Customer Analytics

Customer Lifetime Value

How can the future value of each customer relationship be estimated to inform decisions?

Download PDF

A complementary visual resource for a deeper view of the case.

Relevant industries

  • Retail
  • Banking
  • Telecom
  • Insurance
  • SaaS

Approaches

  • CLV
  • Probabilistic Models
  • Forecasting
  • Value

The business challenge

What economic value might a future relationship generate, with what uncertainty, and for which decisions is it sufficiently reliable?

CLV connects acquisition, retention and service with expected value. A number without a horizon, costs or uncertainty can lead to poor priorities.

Key business questions

  • Which future value should be estimated, and over what horizon?
  • Which revenue, margin, cost and tenure components matter?
  • Which decisions will use the estimate?

Data & evidence

  • Purchases, revenue, margin and returns.
  • Frequency, recency and tenure.
  • Service, acquisition and incentive costs.
  • Product, channel and cohorts.

How I would approach it

  1. Define horizon, economic unit and use.
  2. Separate frequency, value and tenure.
  3. Validate on future cohorts and periods.
  4. Represent uncertainty and sensitivity.

What you would get

  • CLV estimate with explicit horizon and assumptions.
  • Customer value distribution and profiles.
  • Drivers, uncertainty and scenarios.
  • Recommendations for acquisition, service and retention use.

From analysis to action

Illustrative example: CLV is an expectation conditional on data and assumptions, not a revenue promise.

  • Adjust acquisition and service thresholds.
  • Prioritise strategies by value and need.
  • Inform segmentation and planning.

How success should be measured

  • Error and calibration by horizon.
  • Ranking stability.
  • Incremental value of decisions informed by CLV.

Other approaches to consider

Historical, BG/NBD, survival or supervised models may be appropriate. Complexity should be justified by a better decision.