Should loyalty points expire? Expire the reward instead

Loyalty points shouldn't expire on a calendar, but a reward the customer has already earned should, because a short deadline is what gets it redeemed.

Andrew Kim

Photograph by Ashkan Forouzani on Unsplash

Chipotle runs two clocks and publishes both. Its rewards terms, last updated on 13 April 2026, say a reward expires if it isn't used within 60 days of landing in your account, and that points expire after 365 days of account inactivity. One clock is short. The other is six times longer and only starts running if you stop turning up.

So when you ask whether loyalty points should expire, you're asking two questions with opposite answers. Don't put a deadline on the balance. Put one on the reward, because that is the only form of expiry anybody has bothered to measure, and the measurement says it works.

The advice you'll find splits neatly into a liability argument for expiry and a goodwill argument against it, and the two sides have been having that fight for as long as anybody has issued points. Both are about the wrong clock.

The liability you were told to worry about is an accounting rule

Every page I read recommending expiry says the same thing about the balance sheet, and the balance sheet in question belongs to a company that files with the SEC.

Under ASC 606, points a customer earns count as a "material right", which means part of the money already in your till belongs to a future free coffee rather than to the coffee you just sold. BDO works the standard's own Example 52 through: sell $100,000 of food, issue 10,000 points worth $1 each, expect 95% of them back, and $8,676 of that $100,000 stops being revenue until somebody redeems or the points lapse. Expiry releases the deferral. That is what the liability argument is actually about.

The size of it shows up in the filings. Chipotle's most recent quarterly report puts the Chipotle Rewards liability at $80.6 million on 30 June 2026, after deferring $60.2 million of revenue and recognising $47.0 million in those three months alone. That needs managing. Expiry is one of the few levers that moves it without renegotiating with each of the 21 million members one at a time.

Now check whether you have the same problem. If you report on the cash method, which IRS Publication 334 describes as including income when you actually or constructively receive it and deducting expenses in the year you pay them, the $4 you took for that coffee was income in the month you took it. All of it. None of it was deferred against a stamp, so expiring the stamp releases nothing, because nothing was ever set aside to release.

What you do have is a cost that arrives later: the drink itself, whenever somebody claims it. That's a pricing question. It's worth doing that sum properly before you reach for a deadline as the fix, because what the reward costs you to hand over decides how much an unclaimed one was ever worth worrying about.

Nobody is coming for your unclaimed stamps

Federal law does set a floor under gift cards. Under Regulation E's gift card rules, the money on a gift certificate or store gift card generally can't expire sooner than five years after it was loaded.

Loyalty is carved out of that. A "loyalty, award, or promotional gift card" is excluded from those definitions, so the five-year floor never reaches your points. The carve-out costs you a disclosure instead: hand out something prepaid with a stated value under a rewards program, and the front of it has to say it was issued for loyalty, award, or promotional purposes and state when the underlying funds expire.

Unclaimed property law lands in the same place. The 2016 Revised Uniform Unclaimed Property Act defines a loyalty card as "a record given without direct monetary consideration under an award, reward, benefit, loyalty, incentive, rebate, or promotional program", then excludes it from the property a state can claim as abandoned. Eversheds Sutherland, reviewing how the states took up the model, reports that every state revising its escheat standards since 2016 wrote in an express exemption for loyalty cards.

So the rule you print on the card is the only rule there is. Nobody supplies a default. That cuts both ways, because a program with nothing written down has promised in effect that a stamp from 2023 is still good, and sooner or later somebody turns up to test it.

A deadline on the reward is the only expiry with evidence behind it

Suzanne Shu and Ayelet Gneezy handed 64 people a gift certificate to a good French pastry shop and changed one thing about it. Just the deadline. Half had three weeks to use theirs and half had two months.

Ten of the 32 people holding a three-week certificate redeemed it, which is 31%. Two of the 32 holding a two-month certificate did, which is 6%. Their paper in the Journal of Marketing Research reports that gap as significant at p = .01, and their follow-up survey found the people who let the certificate lapse agreeing most strongly with "I was too busy and ran out of time" and "I kept thinking I could do it later", and barely agreeing with the options about forgetting, too much effort, or not liking pastries.

Then there's what a separate group of 80 people predicted before any of it happened. Shown the same two certificates, they expected 68% redemption on the two-month version against 50% on the three-week one. They had it exactly backwards. A longer window feels like generosity and behaves like an invitation to put the thing off, which is why the instinct to be lenient with a reward quietly costs you the visit it was supposed to buy.

The same pair ran it again with free cinema tickets, varying the deadline between two weeks and six while also changing how big the prize was and how far the cinema sat from campus. Redemption came in at 49% short and 35% long. Smaller gap, same direction.

A deadline on an earned reward isn't a trick for clearing a liability, then. It does the reverse. It turns a reward you have already promised into somebody walking back through the door, which was the entire reason for granting it.

A stockpiled point is a customer who isn't coming in

Points sitting in an account do nothing. Chen Li, Srinivasan Swaminathan and Junhee Kim, writing in the Journal of Service Research in 2025, found that customers "tend to stockpile rather than redeem points", which is the academic way of describing the card in the back of somebody's wallet with four stamps on it.

Antavo's Global Customer Loyalty Report 2026, a survey of more than 1,000 US consumers within a global sample of 10,000, measures both ends of that. It found 26.2% of points go unspent and 11.9% expire unspent, while 41.1% of consumers name rewards expiring before they could be used as a frustration and 49.1% say rewards take too long to earn. Those two complaints point in the same direction. People resent losing a balance they were still working toward, and they're fairly relaxed about a deadline on something they've already won, so the deadline belongs at the end where the winning happened.

What the big programs actually expire

Everybody quotes Starbucks on expiry, usually to argue that expiry is normal and that a shop with one till is therefore entitled to do the same. The terms say something more specific.

Program	The points balance	A reward already earned
Chipotle Rewards	Expires after 365 days of account inactivity	Expires 60 days after it lands in the account
Starbucks Rewards, Green	Stars expire six months after the month they were earned, pushed out a month at a time by a purchase, a redemption, or a $30 reload	No general deadline stated; the birthday reward is good on the birthday itself
Starbucks Rewards, Gold and Reserve	Stars don't expire while the level is held	No general deadline stated; the birthday reward runs 7 days at Gold and 30 at Reserve
American AAdvantage	All miles expire if the account sees no qualifying activity in any 24-month period, and any activity resets that for another 24 months	Not offered as a banked reward, since miles are spent at booking
Delta SkyMiles	“Miles do not expire”	Not offered as a banked reward

Every row read off the company's own terms on 2 September 2026: Chipotle, Starbucks, American and Delta.

Look at what the middle column measures. Starbucks' six months isn't really a calendar deadline, because one qualifying purchase in the month before expiry pushes every expiring Star out another month, and you can do that indefinitely. American's 24 months resets on any activity at all. Both are dormancy clocks wearing an expiry date, and Chipotle moved its own the same way in April 2026, extending the previous six-month window to a year while leaving the 60-day deadline on an earned reward exactly where it stood.

Nobody with a pricing team runs a hard calendar expiry on a balance. Not one of them. Several run a tight one on the reward instead, and Starbucks even widens the birthday window as status rises, from a single day at Green to 30 at Reserve.

So should loyalty points expire? Here's the rule to write down

Three lines, and they fit on the back of a card.

A reward, once earned, is good for 30 to 60 days. Print the date on it and say the date out loud when you hand it over.
A balance lapses only after a stretch of nothing so long that no returning regular could hit it. Count it in missed visits rather than months: a café whose regulars come twice a week has watched somebody miss roughly 50 visits by the six-month mark, while a salon seeing clients every six weeks has seen about nine missed appointments in a year, so the café can count its window in months where the salon has to think in years.
Nothing lapses without a warning the customer sees before it happens, not after.

That last line is where programs fail. A stamp card in a drawer cannot tell anybody it's about to lapse, and neither can a spreadsheet of names, which is the one practical advantage a card living on a phone has in this particular argument.

Never sweeping a balance does carry one honest cost, and it isn't on the balance sheet. It's your customer list. Everybody who enrolled and then moved away two years ago still counts as enrolled, which matters because this category tends to price on active customers, and our own free plan caps at 200 of them. Treat the sweep as a records decision rather than a money one, and set it far enough out that a returning regular never meets it.

When a hard expiry is the honest answer

A reward you genuinely can't fund twice changes the calculation. If what sits at the end of the card is a $200 treatment rather than a $4 coffee, an open-ended promise is a real exposure, and a stated expiry is fairer than quietly rewriting the rules two years in. A business with a closed season has a similar problem in reverse: a beach café shut from November is holding balances nobody can spend for five months, so any window has to clear the closure or the promise reads as a trick. And if you've decided to wind a program down, a published run-off date is simply the decent way to do it.

You'll also see it claimed that 56% of customers change how they shop after losing points to an expiry. I can't find the study behind that number, and it turns up on vendor pages citing each other, so treat it as a plausible direction rather than a measurement. Antavo's 41.1% has a survey attached to it.

Settle it before the first stamp. Which reward sits at the end of the card is its own decision, and it moves what an expiry rule costs you far more than the date on it ever will.

Put the clock on the reward

Loyalty Cards keeps the balance and the earned reward on a wallet pass your customer already carries, updated in real time, so nobody discovers at the counter that something lapsed. Starting is free.