Insights

Proving a Loyalty Programme Is Viable Before You Build It

The economics of a coalition programme, and how to model one honestly

Most loyalty programmes are designed as rewards. The good ones are designed as a profit and loss statement.

This piece is about the second – specifically about coalition programmes, where an operator runs a scheme across businesses it does not own, because that is where the economics are least understood and most instructive.

The two rates

A coalition programme is governed by two numbers, and they are deliberately not the same number.

The issue rate. What a participating retailer pays the programme operator for every point issued to a shopper in their store.
The redemption rate. What the programme operator pays a retailer when a shopper redeems points there.

The issue rate is set higher than the redemption rate. The spread between them is the programme’s gross revenue per point, and the points that are never redeemed at all are pure margin on top.

Once you see that, every design decision in every loyalty programme you have ever joined makes sense.

Tiers exist to concentrate the earn rate on high-value members. Expiry rules exist because breakage is modelled rather than regretted. Bonus point promotions are marketing spend drawn against a known unit economic. None of it is arbitrary.

What has to be modelled

A credible five-year model needs four groups of assumptions. Missing any one produces a number that cannot be defended.

1. The traffic and value base

Total sales across the participating estate, and total footfall.
Average transaction value, and visit frequency per customer per year.
Number of outlets participating, which drives both reach and cost.
Number of unique customers – derived, not assumed.

2. The points economics

The issue rate and the redemption rate, and therefore the spread.
Points awarded per unit of currency spent, by tier.
Expired points ratio – earned but never burned.
Points redemption rate per year, which determines how quickly liability converts to cost.
Bonus points issued for marketing purposes, which are cost without corresponding revenue.

3. Adoption, modelled in both directions

This is where most models flatter themselves. A credible one ramps two curves in opposite directions.

Acquisition conversion rising over time as awareness builds and staff engagement improves – and it should be stated explicitly that this depends on stakeholder engagement rather than happening automatically.
Churn rising over the same period. Programmes start with enthusiasts and progressively recruit the less committed. A model with flat or falling churn across five years is not describing a real programme.

4. Operating cost

Hardware at point of sale, per outlet.
Messaging and data costs per member per year.
Programme operations, support and administration.

The gift card is a second business

Modelled separately, with its own revenue lines, its own costs and its own balance sheet treatment.

Denomination mix. Not one card – a range, with a target share for each value. The mix determines average card value and therefore working capital.
Card value as a liability. Money received for goods not yet delivered. It sits on the balance sheet until redeemed.
Issuance fee as revenue. A fixed fee per physical card, typically waived on virtual cards to drive digital adoption.
Breakage. A small proportion never redeemed. Modelled conservatively, and never presented as the business case.
Printing and production cost. Per physical card. The gap between the issuance fee and the printing cost is the unit margin.
Tax treatment. In many jurisdictions the card itself is exempt while the issuance fee is not. Get this wrong and the model is wrong from the first cell.

Test more than one revenue model

The discipline that separates a business case from an advocacy document.

A serious model does not test one structure eight ways. It tests two fundamentally different structures against each other.

Commission per point. Revenue tied to points issued. Predictable, scales with engagement, and requires retailers to accept a per-point charge.
Commission on sales. Revenue as a percentage of transaction value. Simpler for retailers to understand, but decouples your revenue from programme engagement entirely.

Run both. Run multiple scenarios inside each, varying conversion, churn, redemption and cost. Then build the cash flow for the scenario you intend to recommend – because a programme that is profitable in year three and insolvent in year one is not viable.

The output of that work is not a number. It is a defensible recommendation with the alternatives visibly considered, which is the only form of business case that survives a board.

The commercial argument you must win first

None of the modelling matters until retailers agree to participate. They will ask one question.

Why would I give points to shoppers so they can redeem them in a competitor’s store two doors down?

It is a fair question, and the programme does not launch without a convincing answer. There are four parts.

Reciprocity. Every participating retailer issues points and every one accepts them. Your customers spend elsewhere; their customers spend with you. The exposure is symmetrical by design.
Reach. The points you issue go overwhelmingly to people already buying from you. The opportunity is the large majority of the centre’s shoppers who have never entered your store and now have a reason to.
Economics. Acquiring a new customer costs substantially more than retaining an existing one. A mechanism that does both at marginal cost is cheap by comparison.
Data. You learn what your customers buy when they are not buying from you. Nothing else in your marketing provides that.

What to measure once it launches

Set these before launch, because retrofitting measurement to a live programme is close to impossible.

Average transaction value uplift among members against non-members.
Visit frequency, same comparison.
Points earned against points redeemed, tracked as a ratio over time.
Offer redemption and reward redemption rates.
Active members as a proportion of total customers – the number that tells you whether this is a programme or a database.

The first principle, restated

Everything above serves one test.

Programme investment plus promotional cost must be less than the additional customer profit generated.

A loyalty programme is the cost of learning about your customers. If the knowledge is worth less than the programme, it is not loyalty. It is a permanent discount that has become politically impossible to withdraw.

← All insights