A loyalty programme that survives a margin review

The meeting is scheduled for forty minutes and the loyalty programme is item four. By the time it comes up, two other budget lines have been cut, and the finance director opens with a question that sounds neutral and is not. What does the programme cost us per year, all in, including the points people have not spent yet.

Most CRM teams cannot answer that in the room. They can describe member engagement, tier distribution and redemption rates. They cannot state the liability, the margin given away per transaction, or what the programme changed in customer behaviour that would not have happened anyway.

That gap is why loyalty programmes get trimmed rather than improved. A programme that can answer those three questions tends to survive, even when its numbers are mediocre, because it is legible. This piece is about making yours legible before somebody asks.

Know the liability before you argue about engagement

Points and credits are not a marketing metric, they are an obligation. Under the revenue recognition rules most companies now follow, points issued with a sale are generally treated as a separate obligation, with part of the transaction value deferred until they are redeemed or expire. Your finance team will have a specific treatment and a breakage assumption, and you should know both by name.

Ask for four figures and keep them current:

• Points outstanding at the end of the last period, in units and in deferred value.
• The rate at which the balance is growing or shrinking.
• The breakage assumption in use, and when it was last revisited.
• Redemption cost recognised in the period.

Then find out where the cost lands. In some businesses the redemption is charged to the store or region where it is spent, which creates a quiet incentive for local teams to discourage redemption. If that is happening, the programme has an internal problem that no amount of campaign work will fix.

Knowing these numbers changes the conversation more than any engagement statistic. It signals that the loyalty programme is being run as a commercial instrument rather than a marketing activity, which is how it will be judged regardless.

Answer the question finance is actually asking

The question is never "are members more valuable". Members are always more valuable, because your best customers join loyalty programmes. Reporting that comparison as a result is the fastest way to lose a finance audience, and they will spot it immediately.

The question is what the programme caused. Three ways to get closer to an answer, in ascending order of rigour:

1. Pre and post enrolment comparison within the same customers. Take members and compare their behaviour in the twelve months before joining and after. It controls for the selection problem partially, and it is still contaminated by the fact that people join at a moment of high engagement.

2. Matched cohorts. Find non-members who resemble members on prior spend, tenure, category and channel, and compare forward behaviour. Better, and dependent on the quality of the match, which is usually worse than the analyst claims.

3. A genuine holdout. Withhold the programme, or a specific mechanic within it, from a randomly chosen holdout group and measure the difference. This is the only method that produces an incrementality number you can defend line by line.

A full programme holdout is often politically impossible, and that is a legitimate constraint rather than an excuse. Hold out mechanics instead. Bonus point events, tier-anniversary gifts, birthday rewards and point-expiry reminders can all be withheld from a random slice with very little exposure, and each one is a budget line in its own right.

Do at least one of these per quarter and keep the results in a running log. After a year you will have a defensible view of which parts of the programme earn their cost, and which are habit.

Design earn and burn so the margin is knowable

Many programmes cannot state their cost per transaction because the mechanics were set separately over several years and nobody has multiplied them together.

Do the arithmetic per tier and per category. Take the earn rate, apply the expected redemption rate, apply the value per point at redemption, and express the result as a percentage of the transaction. That number is a discount, and it sits on top of every other discount the customer receives.

Two patterns show up repeatedly when teams run this for the first time. The effective discount is higher than anyone assumed, because promotional multiplier events were never included in the baseline. And it is highest in exactly the categories where gross margin is lowest, because the earn rate is flat across a catalogue whose margin is not.

Fixing the second is usually the single largest improvement available. Vary earn by category in line with margin, or cap it in the thin ones. Customers rarely object, because the rule is simple to state, and the finance conversation changes completely once the programme cannot lose money on a product line.

Redemption design matters as much as earn. Redemption thresholds set very high look efficient because breakage rises, but they also mean many members never experience the benefit, which is what actually drives the behaviour you want. A member who redeems something small early is more likely to still be here next year than one carrying a balance towards a reward they will never reach.

Move value towards things that cost less than they are worth

The best answer to a margin review is not a smaller discount, it is a different kind of reward.

Rewards that carry marginal cost well below perceived value are the ones to build the programme around:

• Free or faster delivery and free returns, where you control the rate and the perceived value is high.
• Early access to limited stock, releases or sale periods, which costs nothing and moves sell-through forward.
• Service benefits: priority support, extended warranty registration, easier exchanges.
• Recognition and status that genuinely changes the experience rather than appearing only on a dashboard.
• Access to products or sizes that are hard to obtain elsewhere.

Discount is the simplest reward and the one with the worst economics. It also trains price sensitivity in a population you have specifically selected for loyalty, which is a strange thing to spend money on.

Test the swap rather than assuming it. Offer one cohort a points multiplier and another a service benefit of comparable cost, and compare repeat purchase rate over the following quarter. Run it in one category first.

Report on one page, in their units

When the review comes, bring a single page. Liability and its direction of travel. Effective discount per transaction by tier. Incremental value from the most recent holdout, with the method stated. Twelve-month customer lifetime value for member and non-member cohorts matched on prior spend, with the caveat written on the page rather than left for someone to raise.

Add one forward commitment with a date attached. The annual retention plan is the natural place to put it, and the structure we use for that is in the January retention plan.

The part that stays difficult is timing. Liability accrues immediately and behaviour change shows up over quarters, so any honest report in year one shows a cost with an incomplete benefit. The only way through that is to have said so in advance, with the measurement dates agreed before the programme launched. If yours is already running without those dates, set them now for the next mechanic you change, and stop the cycle there.