The currency you print at the till

Every time a sale credits points, the loyalty program issues currency: small, colourful, carrying your logo — but currency. The customer accumulates it, the balance sheet should record it, and sooner or later part of it will come back to the till to be exchanged for discounts, rewards or products. The program's cost accrues at the sale that issues the points, not at the redemption that extinguishes them: whoever records it only when customers redeem is pushing into the future a cost that belongs to the present.

The count has three multiplied ingredients: how many points you issue per unit of spend, what each point is worth at redemption, and what share of the issued value customers will actually redeem. The product is the cost rate — the percentage of revenue the program withholds on every sale that generates it: in the reference case, 0.70%, that is 14,000 a year on two million of spend.

This calculator does that count and compares it with what you accrue today: the coverage index says whether your program is honestly priced into the margin, or whether you are accumulating a silent debt toward your best customers — because the points, by definition, are owned precisely by them.

Breakage is not a gift: who decides the redemption rate

In the reference, the points issued in a year are worth 20,000 at face value, but the expected cost is 14,000: the difference is breakage — the points that will expire, be forgotten or stay below the redemption threshold. It is the program's only free part, and also its most misunderstood: it is not your choice, it is their behaviour.

Here lives the success paradox: everything that makes the program better — a smoother app, more desirable rewards, expiry reminders — raises the redemption rate, and the cost rate with it. A program going from 55% to 85% redemption costs 55% more, all else equal: marketing success is, in accounting terms, an increase in the liability. It is not a reason to sabotage the program — it is the reason to price it properly before it works.

And there is the stern corollary: the redemption rate estimated today is not tomorrow's. Young programs redeem little (balances have not yet reached the thresholds), mature ones much more — and whoever calibrates the accrual on the first year of life is preparing a surprise proportional to their own growth.

The same program, five levels of belief

The table holds generosity and volume fixed — 1 point per unit of spend, value 0.01, two million of spend — and moves only the redemption rate. The program is the same in every row: what changes is how much the customers believe in it.

The linearity is the news: every extra percentage point of redemption costs 200 a year at these volumes, and the distance between an ignored program and a loved one — from 40% to 95% — is more than a doubling of the cost. It is the number to have in front of you when designing the «redeem your points» campaign: it will work, and it will cost exactly this.

The table also serves as a sensitivity test: if you cannot estimate redemption, compute the cost at your industry's plausible extremes and watch how much the rate and the accrual swing. If the spread is wider than the margin the program is supposed to generate, the priority is not marketing: it is measuring r.

How this calculator works

The annual spend generating points is the revenue on which the program credits: if some channels or products are excluded, only the issuing part counts. The points per unit of spend and the redemption value per point describe the generosity: together they give the face value issued per unit sold — 1 point worth 0.01 means 1% of the price returned in point-currency.

The expected redemption rate is the share of issued value that will actually be redeemed before expiring: the hardest and most important parameter, to be estimated by cohort on the program's historical data. The current accrual is the percentage of spend you set aside today for the points: zero is allowed, and it is the most instructive case.

From there come the expected annual cost, the rate, the shortfall and the coverage index ρ — the dimensionless verdict that works identically for the café with a stamp card and the chain with an app.

How to read the results

The expected annual cost is what the program subtracts from the margin of the year in which the points are born: the number to put next to the benefits (returns, frequency, data) when judging the program. The cost rate is the same number in points of revenue: the part of the price list you have promised to give back.

The annual shortfall compares needed and accrued: positive is today's margin inflated at tomorrow's expense, negative is excess caution. The coverage index ρ = a/w delivers the verdict: below 0.5 the liability is uncovered; between 0.5 and 1 you cover part but not all; between 1 and 1.5 the program is honestly priced; beyond 1.5 you are over-accruing — or overestimating redemption.

The industrial twin

This page has a factory twin: the warranty reserve calculator. The structure is identical — a promise included in the price, a statistical incidence, a rate to accrue at every sale — and only the nature of the promise changes: there «we will repair if it breaks», here «we will discount if you come back».

The parallel teaches in both directions. From the warranty, the loyalty program inherits the accounting discipline: the obligation is born at the sale, and accounting standards treat points as deferred revenue to be recognized at redemption. From loyalty, the warranty inherits the success paradox: there too, making customer service easier raises the claim rate — and the cost.

The family is wider than two members: discount vouchers issued, SLA credits promised, gift cards sold — wherever a promise with measurable incidence travels inside a price, the rate w and the coverage ρ are the right way to count it.

Why the program is worth it anyway

This page measures the cost side, and it is worth saying: a well-designed program pays for it. The mechanism is the repeat purchase rate: if the points move even a few percentage points of customers from single purchase to repetition, the incremental margin usually exceeds the rate — which in the reference is less than one point of revenue.

But the honest comparison requires both numbers, and this is where many programs live on faith: the cost is certain and grows with success, the benefit must be measured with cohorts and control groups — would the program's customers have bought anyway? The difference between a program and a charity is exactly this measurement.

The calculator provides the judgement's denominator: what it costs. The numerator — what it returns — lives in your repeat data, and the practical rule is merciless: if you cannot say by how many percentage points the program lifts repeat purchase, you do not know whether it pays. The deferred discount, meanwhile, keeps running.

Where the numbers come from

The redemption rate is measured by issuance cohort: of the points issued in a given quarter, how many were redeemed before expiry? Programs with history have this number in their data; new programs borrow it from the industry, knowing that redemptions mature over time — and that the expiry policy governs it: eternal points redeem more than 12-month points.

The value per point must be counted at your cost, not at the reward's list price: if 1,000 points buy a product that costs you 7, the value is 0.007 — and that is how well-designed programs offer high perception at controlled cost. If the rewards are discounts, face value and cost coincide.

The current accrual is on the balance sheet or does not exist: many mid-sized companies discover with this count that they have never accrued anything — and that the «points inventory» accumulated over the years is a debt already matured, to be digested before even discussing the current rate.

Limits of the estimate

The model is stationary: it uses an average redemption rate, while real redemption matures over time and reacts to your own moves — campaigns, expiries, new thresholds. On young or fast-growing programs, the steady-state rate is higher than the observed one, and the count must be redone at every cohort measurement.

The cost estimated here is the redemption value: it includes neither platform costs (app, management, reward logistics) nor cannibalization — the share of discounts redeemed on purchases that would have happened anyway at full price. And the benefit is not modelled at all: this is the passive half of the program's income statement, by construction.

A parametric estimate for information purposes: it supports the sizing of the accrual and the comparison of generosity scenarios; it replaces neither the applicable accounting standards — which treat points as deferred revenue — nor your auditor's judgement on the liability to book.

The generosity ladder

The first table holds redemption at 70% and moves the redemption value per point: from the prudent card to the aggressive program giving back 2% of the price.

Generosity is the only lever entirely in your hands, and also the most dangerous to touch: it only goes up. Reducing the value of points already promised is the move programs pay dearest in reputation — the industry's chronicles are full of badly announced devaluations — so the initial value should be chosen the way a list price is chosen: with the intention of keeping it.

The 1.05% row deserves attention: it is where the program crosses one point of revenue and changes category — from marketing cost to price lever. At that weight, every generosity decision should be taken at the same table that decides list discounts, because economically it is the same decision under another name.

The coverage ladder

The second table photographs five accrual policies on the same program — required rate 0.70% — from accruing nothing to accruing double.

The first row is the most widespread among mid-sized companies, and its reading is identical to the warranty's: no accrual does not mean no cost — it means 14,000 a year arriving with no date and no accounting line, concentrated in the periods when customers redeem most. Which are, by retail's law of Murphy, exactly the peak periods.

The last row has a double diagnosis: either the caution is excessive, or — more often — the redemption rate used for the accrual is the marketing department's optimistic one instead of the cohort-measured one. Coverage above 1.5 is an invitation to reconcile the two numbers, not to celebrate solidity.

The points inventory: the debt you already have

This page's rate measures the flow — how much liability you generate this year. But programs a few years old also have a stock: the points issued and never redeemed, accumulated season after season. Its value is outstanding points balance × value per point × residual redemption rate — and in many companies it is a multiple of the annual cost.

The stock has an unpleasant property: it is debt already matured. It can no longer be priced into the sales that generated it — those are closed, the margins distributed — and when it surfaces (a redemption campaign, a platform migration reminding everyone of forgotten balances) it will hit the income statement of the wrong year.

Mature management: measure the stock once a year, accrue it gradually if it never was, and govern its emergence — better a redemption campaign planned in low season than a spontaneous awakening in the peak quarter. The debt does not disappear: you only choose when to pay it.

Designing generosity: high perception, controlled cost

Well-designed programs separate what the customer perceives from what the company pays. The main lever is the reward at cost: 1,000 points buying a product perceived at 12 but paid 7 make a program perceived at 1.2% of the price which, at 70% redemption, costs 0.49% — the spread between perception and cost is free generosity.

The second lever is thresholds: the 500-point reward redeems far more often than the 5,000-point one, and the choice of thresholds governs both r and visit frequency. The third is tiers: concentrating generosity on high-frequency customers buys repetition where the propensity already exists — and that is where incremental margin beats the rate most easily.

All three levers carry the same warning: they move r, and r moves the cost. The calculator should be rerun at every redesign — ten seconds — because the industry's history is full of programs redesigned «for the customer» that doubled the liability without anyone redoing the count.

Expiries and frictions: governing redemption without betraying customers

Point expiry is the most tempting lever and the most delicate. Economically it works: shortening the points' life lowers r and therefore the rate. Commercially it is a pre-announced boomerang: the customer who watches a balance they cared about expire does not lose the points — they lose the trust, and the program loses the reason it existed.

The sustainable version is inactivity expiry: points live as long as the customer is active, and die only with the relationship. It aligns the incentive (come back and your points live) with the economics (breakage concentrates on customers lost anyway) — which is why it has become the standard of mature programs.

Frictions — convoluted redemptions, hidden catalogues, unreachable thresholds — deserve a single word: they are breakage bought with reputation. They lower r today and the sign-up rate tomorrow; this page's count records them as savings, next year's income statement will record them as churn.

Mistakes that sink the count

The first is recording the cost at redemption: it inflates the margin of the growth years — when more points are issued than redeemed — and presents the bill exactly when growth slows and redemptions catch up with issuances. It is the mistake accounting standards exist to prevent, and management accounts commit anyway.

The second is the first year's redemption rate used as the steady-state rate: young balances have not reached the thresholds, and mature redemption can be double the initial one. The third is the reward's list value instead of its cost: it doubles the apparent rate and leads to underestimating the perception-cost spread, that is, the design's best lever.

The fourth is ignoring the stock: calibrating the accrual on the current flow while years of never-recognized points sleep in the cellar. The antidote is the annual inventory of the liability — balance, expected residual redemption, value — next to the current rate: two numbers, once a year, and the program stops being a faith to become an account.

Frequently asked questions

How much does a loyalty program really cost?

The cost rate is w = points per unit of spend × value per point × redemption rate: with 1 point worth 0.01 and 70% redemption, 0.70% of point-issuing revenue — 14,000 a year on two million. Plus platform costs, which this count does not include.

When does the points cost accrue?

At the sale issuing them, not at redemption: from that moment the obligation exists and accounting standards treat points as deferred revenue. Recording the cost only at redemption inflates today's margin at the expense of future years.

What is breakage?

The share of issued value that will never be redeemed: expired, forgotten or below-threshold points. It is the program's only free part — and you do not decide it: it depends on customer behaviour and the expiry policy.

What is the success paradox?

Everything that improves the program — a smooth app, desirable rewards, reminders — raises the redemption rate and therefore the cost. A program going from 55% to 85% redemption costs 55% more: marketing success is, in accounting terms, an increase in the liability.

How do I estimate the redemption rate?

By issuance cohort: of the points issued in a quarter, how many were redeemed before expiry. Young programs redeem little because balances have not reached the thresholds: the steady-state rate is higher than the first year's.

Which value per point: list price or cost?

Your cost: if 1,000 points buy a reward that costs you 7, the value is 0.007 — not the reward's list price. If the rewards are discounts, face value and cost coincide.

What is the coverage index?

ρ = current accrual / required rate. Below 0.5 the liability is uncovered; between 0.5 and 1 you cover part of it; between 1 and 1.5 the program is honestly priced; beyond 1.5 you are over-accruing or overestimating redemption.

What if I have never accrued anything?

The calculation with zero accrual shows the full cost you are pushing onto the future. Mind the stock too: the points accumulated over the years are a debt already matured, to be estimated and digested on top of the current rate.

Are expiring points a revenue?

In accounting, expected breakage reduces the liability to book, and the standards build it into the estimate. But treating expiries as profit is short-sighted: aggressive expiries lower the cost and the perceived value together — that is, the reason the program exists.

Is a loyalty program worth it?

This tool gives the cost side; the benefit must be measured on the repeat purchase rate with cohorts and control groups. The rule: if you cannot say by how many percentage points the program lifts repetition, you do not know whether it pays.

Does it apply to cashback and gift cards too?

Cashback is the case with near-100% redemption (rate = face value); sold gift cards are the relative with upfront collection and their own breakage. The mathematics of rate and coverage is the same family.

Does this count replace the financial statements?

No: it is the managerial sizing of the liability for information purposes. The applicable accounting standards treat points as deferred revenue with precise rules, and the liability to book is judged by your auditor.

Updated July 2026