← All Labs

Customer Lifetime Value

Audience
01

Customer question

Which customers are actually worth acquiring?

02

What this solution does

A cheap customer who buys once isn't necessarily better than an expensive customer who buys for years — this shows the full picture, not just the first sale.

Specifically, it answers: Beyond the first purchase, what is a customer likely to be worth over time, and does that change which acquisition channels or segments are actually the best investment?

03

How it works

Historical customer purchase and retention behaviour is used to project forward value (e.g. 12- or 24-month predicted value), which is then compared against acquisition cost to reveal which acquisition strategies are genuinely profitable.

04

Data requirements

  • Customer acquisition cost by channel or campaign
  • First purchase value
  • Purchase frequency and recency over time
  • Retention/churn indicators
  • Sufficient historical depth to observe repeat behaviour
05

Discovery questions

How much purchase history exists per customer cohort?

CLV projections need enough repeat-purchase history to be credible rather than purely assumed.

Is CAC tracked accurately by channel or campaign?

Without accurate CAC, comparing predicted value against acquisition cost is meaningless.

What decision will this CLV analysis actually inform — budget allocation, channel mix, retention investment?

Shapes whether the model needs to be channel-level, segment-level, or individual-level.

06

Build an experiment

07

Generate synthetic data

08

Data readiness

Generate a dataset in the previous step to run readiness checks.
09

Run model / simulation

Generate a dataset first to run the model.
10

Results

Run the model in the previous step to see results here.
11–12

Business ⟷ Technical interpretation

Customers acquired through highValueChannel show a lower first purchase value but a valueDifferencex higher predicted predictionHorizon value than customers acquired through lowValueChannel — short-term ROAS alone would have favoured the wrong channel.

Shown as a template — the highlighted tokens are filled in once the model has been run.

13

What would you do next?

  • Reweight acquisition channel investment based on predicted value, not just first-purchase ROAS
  • Validate CLV projections against actual realised value as cohorts mature
14

Solution & vendor landscape

Implementation approaches

Internal data science / analytics teamCDP or analytics platform with built-in CLV modellingMarketing analytics consultancy
See the full Vendor Landscape →

Related concepts