Proprietary framework
What is a customer actually worth?
Most businesses know their revenue per customer. Far fewer know what actually drives it, or where the greatest untapped value is sitting.
Over many years and dozens of engagements, I've developed a consistent way of answering that question. It isn't software or a black box. It's a way of looking at customer data most businesses already hold, just haven't looked at properly.
Most customer value work stops at revenue, who spends the most. That misses the real picture. Two customers spending the same amount can be worth very different amounts once the cost to serve them is factored in. Every stage of this model is built on margin, not revenue, because margin is what actually funds the business.
How the model works
- 01
Decile margin mapping
Every customer is ranked by the margin they generate, not the revenue, and split into ten equal bands, not four quartiles, not a rough top-and-bottom cut. That level of granularity shows exactly where value concentrates, and how steeply. The model doesn't assume an 80/20 pattern exists. Sometimes it does, dramatically. Sometimes value is spread more evenly than expected but still meaningfully uneven. The decile view shows which one you're dealing with, rather than assuming.
- 02
The value tier pyramid
Ten deciles are precise but hard to act on in a boardroom. The next step collapses them into three or four tiers a leadership team can actually use, each with its own margin profile: how much margin it contributes, how often those customers buy, and what differentiates them behaviourally, not just demographically.
- 03
The value-by-engagement matrix
The pyramid shows where value sits today. This shows where it's heading. Customer value is plotted against a second dimension, most often purchase frequency, but it can be tenure, lifestage, category breadth or channel behaviour depending on what actually drives the business. That produces a visible trajectory: the path a customer travels from casual to habitual to advocate, rather than a single static snapshot. Customer value here means margin contribution, not revenue, the same distinction that runs through the whole model.
Illustrative. Axes and stages are shaped around each business's own data, this shows the pattern, not any specific client's numbers. - 04
A staged approach, not a single campaign
The model translates directly into a differentiated strategy tied to where a customer sits on that path: build the habit early, in the first weeks or months a shopping pattern is being set; recognise and reward repeat behaviour at meaningful milestones, not on a generic schedule; and treat the smallest, most valuable tier meaningfully differently, not just with bigger discounts.
- 05
Incremental margin and true ROI
Once the tiers and the trajectory are clear, the model goes a step further: it tracks incremental margin per customer and per segment as customers shift between tiers, not a blended average. That's what turns this into a genuine ROI case rather than a segmentation exercise, a defensible answer to what a specific shift in behaviour is actually worth to the business.
What it helps a leadership team decide
- Which customers to prioritise acquiring.
- Which to invest in retaining.
- Where marketing spend is creating value, and where it's being wasted.
- What a specific shift in customer behaviour is actually worth in margin, and what a realistic, evidence-based ROI case looks like.
Where it's been used
This is the same framework behind the Specialty Foods and Best & Less customer value work in Selected Work, refined and adapted differently each time, because the data and the driver of value are never quite the same twice.
Free tool
Apply the model to your own business
The Customer Value Diagnostic is a practical application of this model. Twenty short questions, about seven minutes, and an immediate, personalised view of where your marketing is creating value and where it's being lost.
Want to know what your customers are worth?
No brief required. Start with the situation you're facing.
