Gift Card Breakage - The money left behind

A gift card begins as a promise: the recipient will choose something later. What happens when later never arrives is a question of customer behavior, accounting estimates, contracts and law. This data investigation follows that money through consumer surveys, company filings and academic evidence, and separates reported results from the scenarios used to explain them.

The core research cutoff is 15 September 2026, with Statista data additions reviewed on 16 September 2026. The investigation reviews 14 companies and includes 24 interactive figures and 36 tables. Survey years and company fiscal years remain visible throughout. Source codes in square brackets refer to the annotated bibliography at the end.

What gift card breakage means

A gift can become revenue without becoming a purchase

A gift card starts as a promise. Someone gives up cash today so another person can choose a purchase later. The complication is the interval between those two moments: value can sit on a card, move into an app, be spent in pieces, be lost, or remain usable long after an accountant has recognized some of it as revenue.

This investigation asks four questions: how much value remains unused; how much companies actually recognize as breakage; what evidence supports the estimates; and which party is legally entitled to the residual. Those questions need different datasets. Mixing them produces large, memorable numbers that say less than they appear to.

The central finding is a visibility gap. Public filings can reveal substantial recognized amounts, but they do not provide a complete census of permanent consumer loss. Survey balances include money people still intend to spend. Financial statements apply estimates across many generations of cards. Contract terms and local law determine whether an issuer, a program partner, the purchaser, the recipient, or a public unclaimed-property administrator ultimately receives value.

This is document-based reporting. No interviews were conducted and no private transaction ledger was supplied. Corporate statements are presented as their disclosures; survey answers as self-reports; litigation as attributed allegations or settlements; and analytical scenarios as illustrations. Widely circulated headline estimates are treated as research leads and checked against the underlying evidence before being presented as findings.

T01. Five questions that sound similar but measure different things

The vocabulary determines the result

An unused balance is a stock measured at a point in time. Breakage recognition is a flow measured over a period. A breakage rate is a ratio requiring a defined numerator, denominator and observation horizon. None is interchangeable with the others.

The physical or digital card is also distinct from the obligation. Replacing an expired card may leave its funds intact. Recognizing estimated breakage need not cancel the customer’s redemption right. The most useful analysis follows the monetary obligation across systems, rather than treating a card status field as the final economic outcome.

T02. A working dictionary

A $100 card has more than one clock

Consider an illustrative $100 card sold at face value, with no fees, returns or escheat obligation. Its initial sale supplies cash but creates an obligation. If reliable evidence supports $8 of ultimate breakage, expected customer redemption is $92. After $46 has been redeemed, half that expected use has happened; proportional recognition records $4 of breakage. The remaining customer face balance is $54, while the remaining book liability is $50.

That difference is the heart of the story. An accounting estimate can move faster than the legal extinguishment of an individual customer right. It is therefore unsafe to say that recognizing breakage proves that a named card was abandoned.

The example abstracts from discounts, multi-party settlement, costs and tax. It explains a proportional recognition mechanism; it does not tell an issuer which accounting standard or legal treatment applies to a real program. [M1]

T03. Illustrative lifecycle: face balance versus carrying liability

Illustrative, not a forecast. A residual redemption right may survive accounting recognition. [M1]

What consumer surveys reveal

Fewer Americans reported unused value, but holders had larger balances

The strongest recoverable U.S. survey series covers 2021–2024. Prevalence fell from 51% to 43%, an eight-percentage-point difference, while the conditional mean rose from $116 to $244. These are repeated cross-sections, not the same people or the same cards followed over time. [U01] [U02] [U03] [U04]

The underlying questionnaires concern gift cards, vouchers and store credit. The 2024 coverage includes general-purpose cards and airline or food-delivery vouchers. The 2024 median holder balance was $100, far below the mean: a typical-holder narrative based only on $244 would obscure the distribution. [U05]

A useful denominator check is 0.43 × $244 = $104.92 per adult, using rounded published inputs and including adults with no unused value. The $244 figure is neither a per-card average nor an average for every adult. The national extrapolation is the publisher’s estimate; the public release does not supply enough microdata to construct an independent uncertainty interval. No newer original Bankrate survey was located for this study.

The survey gives two different views of the same unused-value stock. Fewer adults reported holding it in 2024 than in 2021, while the average among holders was higher. Neither observation tells us how much an individual recipient ultimately lost: the people, cards and new gifts entering each survey can change. [U01] [U04] [U05]

T04. Original U.S. survey vintages

Nominal USD; cards, vouchers and store credit. Survey estimates, not audited balances or annual breakage. [U01] [U02] [U03] [U04]

T05. How the surveys were conducted

YouGov online adult samples with weighting; nonprobability-panel design. These panel estimates do not support an invented simple-random-sampling margin of error. [U01] [U02] [U03] [U04]

Who holds unused value, and what people say happened to it

Different demographic cuts answer different questions. The generation chart compares dollar balances only among holders. The income chart measures how common unused value is within each income group. Neither alone establishes which group permanently loses the most money.

Statista also publishes the 2024 generation breakdown shown in F05. Its values—$142 for Gen Z, $332 for millennials, $255 for Gen X and $227 for baby boomers—come from the same Bankrate/YouGov survey already cited here. The original release remains the source for the holder-only denominator and survey method; the additional publication does not supply a second estimate of unredeemed money. [ST01] [U04]

Lifetime experiences also differ from the present stock: the 2024 survey recorded expiry, lost cards and business closures. The categories overlap. Their percentages cannot be added, and they are not the fractions of issued value lost each year. [U05]

Plans to redeem are weaker evidence than observed redemption. In the 2022 survey, 45% of holders expected to use all their unused value within a year and 30% most of it. Without following the same instruments, later aggregate surveys cannot tell whether those plans failed: old cards can be spent while new ones arrive. [U02]

T06. Lifetime loss experiences are overlapping responses

Self-reported experience at any time; not annual incidence or dollar-weighted breakage. [U05]

A reported obstacle is not an ultimate breakage rate

Accenture’s 2024 US Holiday Shopping survey adds a different kind of evidence: people’s explanations for leaving a balance unused. In the table published by Statista, 483 respondents with unused balances could select several reasons. Forgetting the card and having access to only a few brands or stores each drew 34%; 25% said the value was too small to bother using. [ST02]

The ranking does not divide unused cards into separate piles. A person could report a small balance, an inconvenient channel and a forgotten card together. The 23% citing technical difficulties and the 23% citing a single redemption channel cannot therefore be added into a 46% failure rate. Nor does a reported obstacle establish that the money was permanently lost.

These are also different questions from Bankrate’s lifetime experiences in T06. Accenture asks a group holding unused balances why value went unspent; Bankrate asks adults whether specified events ever happened. Even the superficially matching lost-card percentages have different populations. The 25% small-value response gives direct evidence that size can matter, but supplies no universal cutoff below $1, $2 or $3. [ST02] [U05]

T07. Intentions for the next twelve months, 2022

Plans, not tracked outcomes. 71% held at least one instrument a year old; that is not 71% of dollar value. [U02]

A newer snapshot still separates waiting from giving up

CivicScience’s January 2025 reporting, reproduced by Statista, distinguishes several unused-card positions among adults who typically buy or receive gift cards. Eleven percent reported many unused cards, 29% a few, and a further 21% selected the answer that they planned to spend them in the near future. Thirty-nine percent reported none. [ST03] [ST07]

The 21% with near-term plans must not be relabeled as a breakage rate, and the other unused-card categories do not establish an intention never to spend. This snapshot also cannot validate or refute the 2022 plans in T07: different respondents, timing, response options and populations are involved. A credible estimate of ultimate non-use still needs to follow a defined set of obligations over time.

International headlines need their original questions

Ireland’s 2025 consumer research is unusually specific about its denominator. Among 649 respondents who had received Christmas-2024 vouchers, 77% said they had used the full value, 22% had not and 1% did not know. The not-full category includes partial spending; calling it never redeemed would alter the finding. Interviews took place in October 2025. [U10]

The Australian and British examples below are useful context, but cannot be combined into a cross-country breakage league table. The studies differ in instrument scope, time horizon, sample and outcome. A pandemic-expiry episode, current unused holdings and lifetime loss are not competing estimates of the same quantity.

The Irish question also marks the boundary of the evidence. A respondent who had spent nearly all a voucher and a respondent who had spent none could both answer that its full value had not been used. The 22% result identifies incomplete redemption among recipients; it cannot reveal the percentage of voucher money still outstanding. [U10]

T08. Selected international evidence

[U10] [U14] [U13] [U11]

What company filings and research show

What public-company filings actually disclose

The table below covers 14 public companies selected for relevant disclosures. Six disclose separate amounts in the reviewed notes; others describe a policy, provide a liability or combine breakage with redemption. This is an evidence map, not a ranking of all issuers or a representative industry sample.

Reported carrying liabilities are not uniformly measured consumer face balances. Some are net of estimated breakage or fees; Starbucks’ roll-forward includes loyalty Stars. Fiscal labels and year-end dates also differ. Large balances beside missing breakage figures should prompt questions, not invented estimates.

Revenue classification can differ too: Walmart places breakage in membership and other income, while several retailers include it in sales. An analysis that searches only the sales line could miss relevant recognition. [C09]

Even apparently similar labels can cover different products. DICK'S discusses gift cards alongside merchandise credits and added acquired Foot Locker liabilities during the year. Walmart separately discusses reward-related income. Keeping those boundaries visible prevents a convenient comparison from becoming a claim about a uniform gift-card market. [C18] [C09]

T09. Selected latest annual disclosures, USD million

Dates are issuer fiscal year-ends. These rows must not be summed into a national market estimate. Missing is not zero. [C01] [C18] [C13] [C14] [C15] [C20] [C08] [C09] [C04] [C06] [C11] [C10] [C07] [C21]

Starbucks reveals both the scale and the denominator problem

Starbucks’ card breakage series combines company-operated and licensed-store amounts. It increased from $144.6m in FY2020 to $222.4m in FY2025, with a decline in FY2024. That is recognized card revenue, not a measured cohort-loss percentage. [C01] [C02] [C03]

The FY2025 roll-forward reports $15,245.8m of additions and a $1,751.7m closing balance, but combines stored-value cards with loyalty Stars. Dividing card breakage by that addition row would mix scopes. The product also permits reloads, so it is not a pure non-reloadable gift-card benchmark. [C01]

T10. A combined roll-forward cannot supply a clean card-load denominator

Arithmetic: 1,718.7 + 15,245.8 - 15,199.5 - 13.3 = 1,751.7. Mixed scope is preserved. [C01]

The income effect of an estimate change is visible in the footnotes

Cheesecake Factory’s FY2025 breakage was $24.5m, including $17.3m of additional recognition from reassessed redemption patterns. A separate $7.9m gift-card inventory write-down partly offset it; the company’s more precise pre-tax gift-card adjustment, net of the inventory write-down, was $9.396m. Compared with $187.285m operating income, that net adjustment is about 5.0%, an analytical comparison rather than a restated profit measure. [C14]

Chipotle reported $27.877m of breakage in 2025, up from $7.760m in 2024 and $1.162m in 2023. Its $20.1m year-over-year increase is disclosed, but is not identified here as an isolated estimate-change amount. The evidence supports a large recognition increase, not an assertion about its entire cause. [C13]

Brinker supplies the other direction: FY2024 breakage declined by $5.4m, which management attributed primarily to an estimate anticipating higher redemption. Its FY2025 audit treated breakage as a critical audit matter. Accounting estimates can reduce as well as increase reported income. [C17] [C16]

The two restaurant examples show why a breakage line needs its surrounding notes. Cheesecake Factory reported additional recognition alongside an inventory charge; Brinker described lower recognition after expecting more redemption. Reading the revenue entry alone would miss either the offsetting cost or the direction of the underlying estimate. [C14] [C17]

T11. What the disclosures do and do not explain

[C14] [C13] [C17] [C12]

A transparent roll-forward still needs careful labels

Brinker separates sales, redemptions and breakage in its roll-forward, giving readers more visibility than a single combined reduction. Even here, the liability is net of breakage and unamortized third-party fees. Its reported sales row should not automatically be treated as gross face issued. [C15]

Darden reports activations and a combined redemption/breakage reduction. Its FY2026 $751.9m reduction cannot be separated into purchases and breakage from that row alone. Target’s $849m addition is even easier to misread: it is current-year cards still unredeemed, net of breakage, not total annual gift-card sales. [C11] [C04]

T12. Brinker gift-card roll-forward, USD million

Closing = opening + reported sales - redemptions - breakage + other; small rounding differences may occur. [C15] [C16] [C17]

T13. Three common extraction errors

[C11] [C04] [C07]

The research record is strongest on disclosure, weaker on universal rates

An original Journal of Accountancy review of 167 available 2006 filings found only eight companies separately reporting a breakage amount. Its disclosure categories overlap; they are not stages in a funnel. That historical snapshot documents opacity, but cannot establish today’s disclosure prevalence. [A03]

Gregory G. Kaufinger and Chris Neuenschwander’s 2025 paper, Exploring Voluntary Disclosure of Gift Card Breakage: Insights From Agency and Signaling Theory, examines 79 U.S. retail and restaurant firms across FY2013–2022: 790 firm-year observations. Table 2 records disclosure of a breakage amount in 173 of 395 observations before FY2018 and 194 of 395 afterward—43.8% versus 49.1%, an increase of 5.3 percentage points. These are reporting frequencies, not redemption rates. [A01]

The frequently repeated 2–4% retail breakage range comes from Audit Analytics’ July 2017 discussion of breakage revenue relative to gift-card sales. It is historically attributable; it is not a current, representative cohort benchmark. [A09]

The widely quoted fifteenfold result comes from a narrower model: 17 firms that changed disclosure behavior, totaling 170 firm-year observations. Table 9 reports an ASC 606 coefficient of 2.716 and an odds ratio of 15.116. Odds are probability divided by one minus probability; a fifteenfold odds ratio is not a fifteenfold probability. This selected-sample association neither establishes managerial intent nor measures consumer losses. [A01]

T14. Reading research without upgrading its claim

[A03] [A01] [A09] [R1]

Psychology can explain spending choices without measuring abandonment

Chelsea Helion and Thomas Gilovich’s 2014 paper, Gift Cards and Mental Accounting: Green-lighting Hedonic Spending, combines hypothetical choices, a laboratory experiment and store transactions. In the experiment’s 38-person analysis, the gift-card group spent 76% on hedonic items versus 47% for cash. Participants had to spend at least $5; this examined purchase composition, not abandonment. [A05]

The same paper followed 332 Cornell campus-store customers who used both gift and credit cards. Their average hedonic spending shares were 52% and 21%, respectively. The comparison holds the customer constant, but payment choice was observational and departments proxy for individual products. It does not measure permanently unredeemed value. [A05]

Spending composition and eventual redemption are separate outcomes. Evidence that a gift card encourages an indulgent purchase cannot establish a universal balance threshold or a deadline after which the money will never be spent. Estimating either requires tracking comparable balances over time, with redemption, refunds and expiry recorded separately.

A new 2026 study finds high non-use in a different product

Yixuan Liu, Hua Zhang and Eric Zou’s February 2026 working paper analyzes 4,616,004 consumers and 96,829 merchants on one Chinese merchant-prepayment platform during 2023–2024. Credits never expire. Approximately 40% of prepaid value remained unused after a year; treating it as permanent non-redemption requires inference beyond the observed follow-up. [A14]

For prepayments more than 12 months old at merchants still active in the final week of 2024, approximately 17% were wholly unspent and 17% fully spent. The remainder was partially used. These are transaction shares, not shares of monetary value. The reported median merchant ROI of 5.5 equals breakage divided by bonus incentives actually paid, minus one; it is not a full operating-profit measure. [A14]

These bonus-bearing credits were bought for personal use, limiting comparisons with U.S. gift cards. [A14]

Measuring breakage and its economics

The wrong denominator can manufacture a trend