Loyalty program breakage, the share of points members are expected never to redeem, lands on the books like a windfall. It is revenue from points the company was paid for and expects never to honor. Read it that way, and the program misses what the number is saying. Breakage is not a windfall. It is a forecast of the members a program is losing.
Three years ago, a member joined a hotel program for a conference in Denver. He kept earning through work trips and a wedding, until his job changed. He still holds 38,000 points and could book a free night tomorrow. But the program’s breakage model, the formula that predicts which points will go unused, has already scored him as unlikely to come back for them. His balance now counts toward breakage, and its value turns into revenue as other members redeem theirs.
No one at the company chose to let him go. The estimate noticed first, and it filed the news under revenue.
Southwest builds its breakage estimate from engagement
Southwest Airlines shows the link in its annual report. Its Rapid Rewards breakage model reads the points each member has earned and spent, along with his tenure. Southwest calls those inputs “typically representative of a Member’s level of engagement,” and it puts a price on the result. A one percentage point change in the estimate would have moved 2025 passenger revenue by about $256 million.
The number that sets Southwest’s breakage is an engagement forecast worth about a quarter of a billion dollars a point. A rising estimate can be the earliest written record that members are drifting away, because the model reads the signals a loyalty team would watch.
The stakes run as high at the hotel companies. Marriott carries a $7.99 billion loyalty liability, and its auditor brought in actuaries, the experts who price long-term obligations, to test the breakage math.
The trouble is that two teams read the same number for different reasons. Finance books breakage and defends the estimate, while the loyalty team shapes the member behavior underneath it. When each reads the figure through its own lens, the program celebrates the margin and misses the members. Loyalty Juggernaut puts the point plainly: “Loyalty is a balance sheet asset. Most brands still manage it like a marketing experiment.” Breakage is where that gap shows first. At this scale, the estimate belongs in the quarterly business review, with the loyalty team’s read beside finance’s. A program expected to work as an enterprise growth engine cannot have its largest accounting estimate read only as margin.
Breakage is the share of points a company expects nobody to redeem
Points go unused for plain reasons. Members forget: in a Deloitte survey of US loyalty members about their favorite programs, 40% of respondents said they sometimes forget to redeem. Points expire under the program’s rules, or the next reward sits out of reach. The catalog of rewards holds nothing he wants, or he has left the program.
For the books, the accounting rules use expected breakage, set when points are issued and updated as members act. Realized breakage, an industry term, counts the points that have expired.
A big balance is not the same thing as breakage. One reason can be healthy: some members save for a bigger reward, and a 2015 study found customers stockpile points even when waiting earns them nothing extra. Points that are saved and later redeemed never become breakage. Counting every live point as lost overstates the number. Take the illustrative vintage in the formula table below, one batch of points issued together: counting live points turns its 20% expected rate into 30%. That gap is where programs fool themselves, writing off the members still saving alongside the members who left.
The balance worth worrying about is the one that stopped moving, so the loyalty team’s first job is to tell the two apart.
The accounting rules turn breakage into an estimate that moves revenue
ASC 606 and IFRS 15, the revenue rules for US and international reporting, both treat a point as a promise the company has already been paid for. Its value waits on the balance sheet as a liability, an amount the company owes, until the member uses it. Accountants call that held-back value deferred revenue: cash received for something not yet delivered. Breakage is the part of that liability the company expects never to pay out.
The rules book expected breakage as revenue “in proportion to the pattern of rights exercised by the customer,” meaning as the rest of the points are redeemed. They also cap the estimate, so a company books only the breakage it is unlikely to have to take back later.
The cap exists because estimates go wrong. In July 2022, Southwest dropped the expiry date on its flight credits, which work like points. Customers used more of them than planned, and in the last quarter of 2024 Southwest reversed $116 million of breakage revenue it had already booked. Since May 28, 2025, new credits expire again.
Change the rules, and the estimate moves with them. That is why breakage is never finance’s problem alone, and why the loyalty team belongs in the room.
Average breakage rates are hard to trace, and the filings price the estimate
The figure that turns up most in searches for an average breakage rate is about 25% for retail programs. The earliest source found for it is a 2012 magazine column that credits unnamed studies, so it would not survive a finance review. A 2011 industry estimate that a third of the value of US points goes unredeemed is a different measure, perceived value across every program at once.
Public filers publish something more useful: what a change in their own estimate is worth, a figure analysts call sensitivity. Southwest, American, Delta, and Marriott each state one in their 2025 annual reports, collected in the table below. Each filer uses its own terms, so no two figures combine into a shared rate.
A 2025 study points the same way: researchers built a breakage model for a large regional airline. Breakage varied widely from customer to customer and from market to market. An industry average would blur the very differences a loyalty team needs to see.
The benchmark that matters is not the industry’s. It is the program’s own, tracked by segment and by market over time.
Devaluation cheapens every point and spends trust
There is a second way to shrink the promise. Devaluation means raising the points a reward costs, so each point buys less and costs the company less to honor. In a liability measured at the cost of future redemptions, like Hilton’s, that shrinks the liability. Delta described the same lever to lenders.
The math works on the balance sheet. It does not work on the member, who notices when the reward he was saving for moves further away.
Breakage members drift into is one thing, and breakage the plan depends on is another. Once the forecast starts carrying the margin, breakage has stopped being an estimate and become the strategy. A points currency holds its value only while members trust it will be there when they reach for it.
Consultants have measured the effect, and regulators have started asking questions. IdeaWorksCompany tracked how far economy award prices at six US airlines have outrun inflation. The US Department of Transportation opened an inquiry into airline rewards. The Consumer Financial Protection Bureau warned card issuers about cutting the value of rewards customers had already earned. A Senate bill would ban expiry dates on airline points and make airlines post a redemption rate. As of July 2026, the bill sits in committee without a cosponsor, and the inquiry has produced no rule.
A posted rate would give members a rough read on how much of the currency goes unused. A loyalty team should know its own redemption rate by segment, and what it says about the program, before any law asks for it.
Design choices decide whether a member can reach a reward
Five design choices push breakage up or down, and each decides whether a member can reach a reward. Expiry is the bluntest: a short clock on every point raises breakage. Expiry tied to inactivity, with a warning well before the date, spares members who are still active.
Reward reach comes next, because a high first threshold leaves small balances stalled below it, while a small reward within reach gives them somewhere to go. The catalog has to hold rewards that fit a member’s week, partner options included. Reminders help when they name a reward he can already afford. The earn rate cuts both ways. A slow rate pushes rewards out of reach, and a generous one raises the bill, which forces a leaner catalog or a higher breakage assumption.
Every one of these rules belongs to the loyalty team, which means the loyalty team is already setting breakage, whether or not it watches the number.
Five decisions keep breakage honest
The job is not to push breakage up or down. It is to read what the number is saying about the members behind it.
- Estimate breakage by segment and by market. A single rate blends savers, lapsed members, and new members into a number that fits none of them. Finance reports a blended number, but the loyalty team manages the groups behind it, so ask finance which segments move the estimate most.
- Separate the savers from the lapsed. A rising balance on an active account is a plan, and a frozen balance on a quiet account is a goodbye in progress. Give the saver a reward worth waiting for and the lapsed member a reason to come back.
- Put a small reward within reach. In the 2025 airline study’s model, promoting more frequent redemptions lowered breakage more than favoring rewards that need more points. A 2014 study of 3,094 members found that choosing to redeem lifted purchases before and after the redemption, an effect the researchers call redemption momentum. Budget each redemption as an investment in the next visit.
- Price every rule change before it ships. An award chart sets how many points each reward costs. A new chart, or a shorter expiry window, changes the program’s economics from day one. The books catch up when the estimate is next revised, so model the change with finance first.
- Read the estimate as an engagement report. A rising estimate can mean members are drifting away. Put breakage on the loyalty dashboard next to active members and redemption rates. When it jumps, check for a rule change first, then treat what remains as churn risk: members likely to leave for good.
GRAVTY® shows the loyalty team where balances went quiet
Inside GRAVTY®, the Denver member’s three years stay on one record. Patented multi-dimensional behavior tracking holds every point he earned and spent, from the Denver trip to the wedding.
Say redemptions in his market drop after a catalog change. AI-Sense, GRAVTY’s AI-powered monitoring, spots the anomaly and alerts the loyalty team. The team then asks Agentic AI Compass, launched in October 2025, what changed. Compass compares his segment with others and drills into the anomaly: balances that went quiet after the change.
Because Compass simulates outcomes before execution, a smaller reward that members like him can reach gets tested before the catalog changes again. The team adjusts the catalog rule in GRAVTY Visual Rules, the patented no-code rules engine, and GRAVTY’s patented mass individualization shapes the offer for him. He gets a note about a free night his points already cover.
What the team gains is time to act while his balance still means something to him.
The best breakage is the kind a member chooses
He books the free night for a family trip in June. His balance shrinks, the liability shrinks with it, and the model moves him back among the members expected to redeem. The program gave up a little breakage and kept a member its own model had already let go.
That is the trade the loyalty team and finance have to make together. Breakage a program earns by being forgotten looks like margin. It behaves like a bill that arrives later, in the shape of a member who stopped coming back.
Members forget points. They remember whether the program remembered them.
The numbers behind loyalty program breakage
These are the figures behind this essay. Each sourced figure is linked in its table, and the worked example is illustrative.
The breakage rate formula, with a worked example
Each rate below divides by the points issued in one illustrative cohort, meaning a batch of points issued over the same period.
| Rate | Formula | Illustrative cohort: 1,000,000 points issued, two years on |
|---|---|---|
| Redemption rate | Points redeemed ÷ points issued | 700,000 redeemed = 70% so far |
| Realized breakage rate | Points expired ÷ points issued | 150,000 expired = 15%. It climbs as the cohort ages. |
| Expected breakage rate, the one the accounting uses | Points expected never to be redeemed ÷ points issued | 150,000 expired + 50,000 of the 150,000 live points expected to go unused = 20% |
| Breakage if every live point were counted | (Points expired + all live points) ÷ points issued | 300,000 = 30%, which overstates breakage |
| Breakage revenue recognized so far | Value of expected breakage × share of expected redemptions made | $2,000 of expected breakage on $10,000 of deferred revenue × 700,000 of the 800,000 points expected to be redeemed (87.5%) = $1,750 |
ASC 606 and IFRS 15 rules, set against the 2025 filings
| Question | What ASC 606 and IFRS 15 require | What the 2025 filings show |
|---|---|---|
| Where does the value of unredeemed points sit? | On the balance sheet, as a liability, until members redeem | Marriott reports a $7.99 billion loyalty liability; its auditor made the program a critical audit matter and used actuaries to test breakage. Hilton reports $2.9 billion of liability and $1.5 billion of deferred revenue, and outside actuaries help it estimate what each point will cost. |
| How is breakage estimated? | As the share of points never used, capped so a large revenue reversal is unlikely | Southwest scores members on tenure and points earned and redeemed |
| When does it become revenue? | As the rest of the points are redeemed, or once redemption is remote if the company cannot estimate it | Southwest books breakage in proportion to the points members use |
| What moves the estimate? | New evidence on member behavior and any change to program rules | Hilton’s cost-per-point estimate includes the devaluation or appreciation of points |
| What does a change do to revenue? | A higher estimate raises revenue, and a lower one moves value back into the liability. Only the timing changes. | Updated breakage and funding estimates cut Marriott’s 2025 revenue by about $102 million |
Four filers price a change in their breakage estimate
Each filer uses its own terms, so no row gives a shared rate.
| Filer | What a change in the breakage estimate is worth | Annual report filed |
|---|---|---|
| Southwest | A one percentage point change would have moved 2025 passenger revenue by about $256 million | February 2026 |
| American | A 10% rise in its estimate of miles never used would have added about $140 million to 2025 revenue | February 2026 |
| Delta | A 10% change in the miles expected to be redeemed would move less than 1% of 2025 operating revenue, against $9.3 billion of SkyMiles deferred revenue | February 2026 |
| Marriott | A one percent cut in its breakage estimate could add about $50 million to its loyalty liability | February 2026 |
Devaluation, and who has questioned it
| Who | When | What it said or did |
|---|---|---|
| Delta, in a deck for lenders | September 2020 | SkyMiles sets the miles each award costs, which the deck called “flexibility to control costs and preserve margins.” |
| IdeaWorksCompany | February 2025 | Economy award prices at six US airlines rose nearly 12 points faster than inflation since 2019, and reward value per dollar spent fell by about half. |
| US Department of Transportation | September 2024 | Opened an inquiry into the rewards programs of the four largest US airlines. Then-Secretary Pete Buttigieg said these rewards are “controlled by a company that can unilaterally change their value.” |
| Consumer Financial Protection Bureau | December 2024 | Circular 2024-07 warned card issuers that cutting the value of rewards customers have already earned can break federal law. |
| Senator Dick Durbin | March 2026 | S. 4244, the Protect Your Points Act of 2026, would ban expiry dates on airline points and make airlines post the share of points members redeemed over the past 12 months. |
Five design levers that move breakage
| Lever | Raises breakage | Lowers breakage |
|---|---|---|
| Expiry | A short clock on every point | Expiry tied to inactivity, with a warning well before the date, for members who are still active |
| Reward reach | A high first threshold that small balances stall below | A small reward within reach |
| Catalog | Rewards members have no use for | Rewards that fit a member’s week, including partner options |
| Reminders | No reminder, so balances get forgotten | A balance reminder that names a reward the member can already afford |
| Earn rate | A slow earn rate that pushes rewards out of reach | A generous earn rate, which also raises the bill. The loyalty program design guide shows how it forces a leaner catalog or a higher breakage assumption. |
Frequently asked questions
What is loyalty program breakage?
Loyalty program breakage is the share of issued points that members never redeem. A company estimates it when points are issued and revises it as members act. If it can estimate breakage reliably, it books that share as revenue as the other points are redeemed. The estimate counts expired points and live points expected to go unused, which is why the loyalty team’s read of member behavior feeds the number finance reports.
How do you calculate a loyalty breakage rate?
A loyalty breakage rate is the points expected never to be redeemed divided by the points issued in the same cohort. That expected rate is the one the accounting uses, and it is the mirror image of the redemption rate. The realized rate divides expired points by all points issued. Count a live balance only for the share expected to go unused, because points still being saved would overstate breakage.
What is the average breakage rate for loyalty programs?
No reliable average exists. The most quoted figure, about 25% for retail programs, comes from a 2012 Loyalty Magazine column that credits unnamed studies. In 2011, COLLOQUY and SWIFT EXCHANGE estimated that at least a third of the perceived value of US points and miles, about $16 billion a year, goes unredeemed. Filers publish sensitivity instead: in its 2025 10-K, Marriott says a one percent drop in its breakage estimate could add about $50 million to its loyalty liability. Benchmark against your own segments.
Is breakage revenue?
Yes: under ASC 606 and IFRS 15, expected breakage becomes revenue in proportion to the points members redeem. Until then, the value of issued points sits on the balance sheet as a liability. When breakage cannot be estimated with confidence, the company waits until redemption is remote. The points liability accounting guide covers the full mechanics.
Is high breakage good or bad for a loyalty program?
High breakage is a bad sign for a loyalty program, even though it lifts reported margin. Unredeemed points come from members who forget or lapse, and from members who never get close to a reward. In a Deloitte survey of 5,564 US adults, 40% said they sometimes forget to redeem. A 2014 study of 3,094 members found that deciding to redeem lifted purchases before and after the redemption. Read breakage by segment, and treat a rising rate among once-active members as early churn.
How is loyalty breakage different from gift card breakage?
Gift card breakage is unspent cash a customer paid for, and loyalty breakage is unredeemed value a program issued as a reward. The accounting is the same. Under ASC 606, expected breakage becomes revenue as the remaining rights are used, and money owed to a state under unclaimed property laws stays a liability. The cost differs. A point’s cost moves when the program changes how many points a reward takes, a factor Hilton’s 2025 10-K builds into its estimate.
Should loyalty points expire?
Loyalty points should expire only after a stretch of inactivity, and only after a clear warning. That design spares members who still earn and clears balances members have abandoned. Starbucks’ US rewards terms, effective March 2026, are a model to study. A Green member’s Stars stay active while his membership stays active, and members can track expiration dates in the app. Blanket expiry draws scrutiny: S. 4244 would ban expiry dates on airline points.
