Arrow
Get your free stagnation score
Arrow
Find Your Solution
Logo

Starbucks Recognized $196 Million From Coffee Nobody Drank

by | Sep 17, 2026

FICTION · An Unfair Advantage Intervention

Fiction. Real history. Sourced below.

In fiscal 2023 Starbucks recognized $196.1 million in revenue from money loaded onto cards that will never be spent. Run the Matrix of Profitability and something odd appears: the loyalty program that serves Starbucks’ best customers is partly funded by its most forgetful ones. The Bear Hug cell is subsidized by the cell every standard analysis writes off.

I’m Samantha Reynolds, and I designed Cartwell’s incentive system — the points tied to project results — so I have a professional interest in one question above all others: does an engineered incentive produce genuine commitment, or does it produce compliance that looks identical right up until it stops? The Starbucks Card is the best-documented experiment in that question anywhere, and the filings have started to answer it.

Full fiction disclosure and sourcing note

This is a work of fiction. Samantha Reynolds is a fictional character from The Unfair Advantage by Todd Hagopian, and Cartwell is the book’s fictional company. Samantha analyzes the real, documented history of the Starbucks Card from the present day; she never worked at Starbucks, never met or spoke with any real person, and no dialogue is invented for anyone. Financial figures are drawn from Starbucks’ own SEC filings and earnings releases where available; figures attributed to outside analysis are identified as such. The historical facts are accurate and sourced as of 2026; the framework analysis is invented.

What the Machine Actually Does

Starbucks launched a reloadable prepaid card in 2001, and it now sits at the center of the customer relationship. The scale is easier to grasp in flow than in stock: Starbucks reported record U.S. card loads of $3.6 billion in a single quarter, ranking second among U.S. brands in holiday gift card activations, with 34.3 million active U.S. Rewards members at that point.

Three mechanisms run off that. Customers prepay, so Starbucks holds their cash before delivering anything. Loaded balances make going elsewhere feel like waste, so people don’t. And card-and-app customers are identified rather than anonymous, which turns every purchase into behavioral data.

Outside analysis puts the stored-value float recently around $1.8 billion, with Rewards members driving somewhere between 41% and 57% of U.S. company-operated revenue and spending roughly 2.5 to 3 times what non-members do. Treat those as reported estimates rather than filings figures, but the direction is not in dispute.

All of that is genuinely impressive and it is the part everyone tells. What interests me is the accounting line that never appears in the celebration, because it changes who is paying for the machine.

The Matrix of Profitability: A Textbook Bear Hug

The 80/20 Matrix of Profitability sorts a business across two axes — the top 20% of customers against the top 20% of products — producing four cells. Q1, the Bear Hug, is top customers buying top products, and it is where a business should concentrate. Q2 is the Scale Opportunity, the other 80% of customers buying top products. Q3 is the Strategic Challenge, top customers buying peripheral products. Q4 is the Value Destroyer, everyone else buying everything else.

Starbucks is one of the cleanest Bear Hug operations in retail. Its top customers are Rewards members; its top product is a small set of core beverages bought habitually. The card and app exist to do three things to that one cell: identify it, deepen it, and hold it. Registering a card enrolls you in Rewards automatically, which converts an anonymous cash transaction into an identified customer record.

The recursion is where it gets interesting. 80/20 applied twice says the top 4% of customers generate around 64% of value, which implies a tiered allocation — roughly 60% of your resources to that top 4%, 30% to the 4-to-20% band, 10% to everyone else. Most companies can’t act on that because they can’t tell the tiers apart. Starbucks can, continuously and algorithmically, which is what personalized offers actually are: recursive 80/20 executed per customer, per day.

So far this is the framework applied and the answer is unambiguous. Starbucks found its Bear Hug and built a machine around it. The interesting question is what pays for the machine.

The Inversion: Who Funds the Bear Hug

Here is the finding, and it’s in the filings rather than in any case study.

Starbucks’ fiscal 2023 10-K reports breakage revenue — money loaded onto cards and never redeemed — of $196.1 million, against $196.0 million in fiscal 2022 and $164.5 million in fiscal 2021. The filing explains the mechanism plainly: U.S. stored value cards carry no expiration date and no service fees that decrement balances, and based on historical redemption patterns a portion is not expected to be redeemed and is recognized as breakage over time.

One honest qualifier the celebratory version and the critical version both skip: the 10-K notes those redemption estimates also account for remittance to government agencies under unclaimed property laws. Not every abandoned balance is kept; some escheats to states. The number Starbucks recognizes is what survives that.

Now ask which Matrix cell that $196 million comes from. Not the Bear Hug. Rewards regulars spend their balances down — that’s the definition of the behavior. Breakage accrues from occasional, disorganized, and gift-receiving customers who load or receive money and lose track of it. In Matrix terms, that’s the bottom-80% cells.

Which inverts the standard reading. A conventional 80/20 analysis treats the Q4 cell as a value destroyer to be de-prioritized. Here it is doing the opposite: contributing roughly $196 million a year of essentially pure-margin revenue that helps fund the program serving Q1. The cell you’d cut is subsidizing the cell you’d feed.

I don’t think that makes Starbucks villainous — no one is required to remember a $4 balance, and the terms are disclosed. But it does mean the honest version of “concentrate on your vital few” has a second half: know which cell is absorbing the cost of that concentration. The Matrix will show you where the revenue is. It will not show you who is paying for the program unless you go and look.

A Switching-Cost Moat, and What the Numbers Show

The Moat Taxonomy names six defensive types — cost, brand, switching cost, network, scale, regulatory — and the card is a deliberate switching-cost moat. Loaded money plus sunk-cost bias makes leaving feel wasteful; friction reduction makes staying easy.

The question worth asking is whether an engineered switching cost, absent genuine preference underneath, behaves like a durable moat. This is the same question I ask about any incentive system: does it produce commitment or compliance? And the recent filings give it an unusually clear answer.

Active U.S. Rewards membership was 34.3 million in the first quarter of fiscal 2024, up 13% year over year. By the fourth quarter it was 33.8 million, up 4% and flat sequentially. In that same quarter global comparable store sales fell 2%, driven by a 4% decline in comparable transactions and partly offset by a 2% increase in average ticket.

Read that signature carefully, because it is exactly what a switching-cost moat looks like under pressure. Spend per visit holds or rises — the locked-in customers keep spending when they come. Visits fall. Membership growth decelerates from 13% to 4% to flat. The moat is doing its job on the people already inside it and doing nothing to bring new ones in.

A brand moat produces preference, which shows up as traffic. A switching-cost moat produces retention, which shows up as ticket. When your transactions decline while your ticket rises, you are learning which of the two you actually have.

The Precondition Nobody Copies

The last correction is about transferability, and it’s the one that costs companies money. The Bear Hug machine compounds here because coffee is a high-frequency, low-ticket, habitual purchase — bought almost daily, cheaply, on autopilot, hundreds of times a year. Prepaid balances accumulate because reloading is routine. Habit forms because repetition is constant. Recursive personalization works because there is enough transaction volume per customer to personalize against.

Attach the same mechanism to an infrequent, high-consideration purchase and none of it happens. The float never builds because nobody preloads for something they buy twice a decade. The routine never forms. The data is too sparse to tier. And the sunk-cost lock-in never bites because there is no habit loop to exploit.

The underlying principles still transfer — reduce friction, collect payment early where you legitimately can, use customer identity to serve rather than only to sell. The specific machine does not, and the Matrix is silent on this, because it tells you where value concentrates without telling you which retention mechanisms your purchase frequency can support.

What Actually Happened

Starbucks launched its reloadable prepaid card in 2001, later pairing it with the Starbucks Rewards program and a mobile app. Customers who register a card are automatically enrolled in Rewards in nearly all markets, and members can also earn Stars paying by cash, card, or mobile wallet, so the physical card is no longer required for participation.

Amounts loaded are recorded as deferred revenue and recognized when redeemed; historically the majority of cards are redeemed within a year. U.S. cards carry no expiration date and no balance-decrementing service fees. Based on historical redemption patterns — and net of remittance to states under unclaimed property laws where applicable — a portion is recognized as breakage: $164.5 million in fiscal 2021, $196.0 million in 2022, and $196.1 million in 2023.

In the first quarter of fiscal 2024 Starbucks reported record U.S. card loads of $3.6 billion and 34.3 million active U.S. Rewards members, up 13% year over year. By the fourth quarter of fiscal 2024 membership was 33.8 million, up 4% and flat sequentially, with global comparable sales down 2% on a 4% transaction decline offset by a 2% ticket increase, and non-GAAP operating margin down 380 basis points.

Outside analysis places the stored-value float recently near $1.8 billion and Rewards members at 41% to 57% of U.S. company-operated revenue, spending roughly 2.5 to 3 times what non-members spend.

From the Operator’s Chair

I’m Todd Hagopian, and float, switching costs, and the difference between retention and preference are things I’ve worked with across Berkshire Hathaway, Illinois Tool Works, Whirlpool, and now at JBT Marel. Three disciplines.

First, when you concentrate on your vital few, find out who absorbs the cost of that concentration. The Matrix is excellent at showing where revenue clusters and silent about who funds the program. At Starbucks the answer is roughly $196 million a year from customers who forgot a balance — the exact cell a standard analysis would deprioritize. That’s not necessarily wrong, but you should know it, because a model partly funded by customers not paying attention is a model with a dependency you haven’t named.

Second, learn to read which moat you have from the traffic-versus-ticket signature. This is the most practically useful thing in the case. Preference shows up as visits. Lock-in shows up as spend per visit. When transactions fall 4% while average ticket rises 2% and membership growth goes from 13% to flat, you are watching a switching-cost moat hold the people inside it while failing to attract anyone new. I’d rather know that from my own numbers two years early than from a strategy offsite two years late.

Third — and this is my own trade — never assume an incentive system is producing commitment. I’ve built points programs, and the seductive thing about them is that compliance and commitment generate identical metrics right up until the moment they diverge. The test isn’t engagement, it’s counterfactual: would this person still choose us with no balance loaded and no points to chase? If you can’t answer that, your retention number is telling you how well the mechanism works, not how much anyone wants what you sell. Engineered retention is real and worth having. It just isn’t the thing it’s usually mistaken for.

The Monday Morning Playbook

  • Run the Matrix, then ask who funds the program. Identify your Bear Hug cell — and separately identify which cell is absorbing the cost of serving it. Revenue concentration and cost concentration are rarely in the same place.
  • Track traffic and ticket separately. Rising ticket with falling transactions is the signature of a switching-cost moat without preference underneath. It looks fine in aggregate revenue for several quarters.
  • Ask the counterfactual question about every loyalty metric. Would this customer choose us with no balance and no points? If nobody can answer, you’re measuring the mechanism, not the relationship.
  • Collect payment early where you legitimately can — and price the customer’s side. Deposits, prepayment, retainers, and subscriptions are real low-cost working capital. Know what the customer gives up, because a model that depends on them never noticing is a countdown.
  • Match the mechanism to purchase frequency before copying it. Prepaid float and habit lock-in compound for frequent, cheap, routine buys and do essentially nothing for infrequent, high-consideration ones.

The Contrarian Reappraisal

The consensus reading — the Starbucks Card is the most elegant loyalty mechanism ever built — is mostly right, which is exactly why the missing parts matter.

The Bear Hug is partly funded by the cell you’d cut. Breakage of $196.1 million in fiscal 2023 comes from occasional and disorganized customers, not from the Rewards regulars who spend their balances down. A standard 80/20 read deprioritizes that cell; here it is subsidizing the program serving the top one. The framework shows where revenue concentrates and not who pays for the concentration.

Much of the “loyalty” is a switching-cost moat, and the filings now show its signature. U.S. Rewards membership went from 13% year-over-year growth to flat within four quarters, while transactions fell 4% and average ticket rose 2%. Locked-in customers spending more per visit while fewer visits occur is retention holding as preference erodes.

The mechanism is narrower than the legend claims. It compounds because coffee is a daily, cheap, habitual purchase with enough transaction volume to personalize against. On infrequent or high-consideration purchases the float never accumulates and the habit loop never forms. The principles transfer; the machine does not.

The Starbucks Card Timeline

  • 2001 — Starbucks launches a reloadable prepaid card, initially a convenience and gifting product.
  • Late 2000s — the card is paired with Starbucks Rewards and a mobile app, turning a payment tool into an identified-customer platform.
  • 2010s — Mobile Order & Pay and deepening analytics make the app the center of the customer relationship; registering a card auto-enrolls customers in Rewards.
  • FY2021 — breakage revenue of $164.5 million recognized from balances not expected to be redeemed.
  • FY2022–FY2023 — breakage of $196.0 million and $196.1 million respectively.
  • Q1 FY2024 — record U.S. card loads of $3.6 billion in a quarter; 34.3 million active U.S. Rewards members, up 13%.
  • Q4 FY2024 — membership 33.8 million, up 4% and flat sequentially; global comparable sales down 2% on a 4% transaction decline against a 2% ticket increase.

By the Numbers

  • $196.1 million — breakage revenue recognized in fiscal 2023, against $196.0 million in 2022 and $164.5 million in 2021, per the 10-K.
  • $3.6 billion — record U.S. card loads in a single quarter (Q1 FY2024), ranking second among U.S. brands in holiday gift card activations.
  • 13% → 4% → flat — the deceleration in active U.S. Rewards membership growth across fiscal 2024.
  • −4% transactions / +2% ticket — the Q4 FY2024 signature of a switching-cost moat holding spend while losing visits.
  • 41–57% of U.S. revenue — the reported share attributed to Rewards members, who are estimated to spend 2.5–3x non-members.
  • No expiration, no decrement fees — the card terms that make breakage a function of forgetting rather than of policy.

Frequently Asked Questions

What is Starbucks breakage and how much is it?

Breakage is money loaded onto stored value cards that is never redeemed and is recognized as revenue. Starbucks’ 10-K reports $196.1 million in fiscal 2023, $196.0 million in 2022, and $164.5 million in 2021. U.S. cards have no expiration date and no fees that decrement balances, so the estimate is based on historical redemption patterns and accounts for remittance to states under unclaimed property laws where applicable.

How does the Starbucks Card make money beyond coffee sales?

Through float and breakage. Customers prepay — Starbucks reported record U.S. card loads of $3.6 billion in a single quarter — giving the company use of customer cash before delivering anything. A portion is never redeemed and becomes essentially pure-margin revenue. The card also converts anonymous cash customers into identified ones, enabling personalized offers.

Is Starbucks Rewards real loyalty or just lock-in?

The recent numbers suggest substantial lock-in. Active U.S. membership growth went from 13% year-over-year in Q1 fiscal 2024 to 4% and flat sequentially by Q4, while global comparable sales fell 2% on a 4% decline in transactions offset by a 2% rise in average ticket. Locked-in customers spending more per visit while total visits fall is the signature of a switching-cost mechanism holding as underlying preference softens.

What does the customer give up with a Starbucks Card?

Use of their money before they consume, any interest it might earn elsewhere, the ability to redeem the balance for cash, and deposit insurance — a closed-loop balance is not a bank deposit. They also trade a continuous behavioral record for convenience and rewards. Many would take that trade knowingly; the point is that it is a bargain with two sides, not a free gift of data.

Can other businesses copy the Starbucks Card model?

Only those built on frequent, cheap, habitual repeat purchases. The float accumulates because reloading is routine, the habit forms because repetition is constant, and personalization works because transaction volume per customer is high. On infrequent, high-consideration purchases none of that happens. The underlying principles — reduce friction, collect payment early where legitimate, use identity to serve — still transfer.

About the Stagnation Assassin

Todd Hagopian is a Fortune 500 transformation executive who has generated $3B+ in shareholder value across Berkshire Hathaway, Illinois Tool Works, Whirlpool, and JBT Marel, where he serves as VP of Product Strategy and Marketing. Known as The Stagnation Assassin, he turns proprietary frameworks — including the 80/20 Matrix of Profitability and the Moat Taxonomy — into field-tested tools for operators who need to know where value concentrates and who is paying for the concentration. He is the author of The Unfair Advantage and the Stagnation Assassin series, a Forbes contributor, and an award-winning keynote speaker. Connect with Todd on LinkedIn or bring him to your stage via his speaking page.

The Matrix tells you where the revenue concentrates. It doesn’t tell you who’s funding the program — and at Starbucks that’s $196 million a year from people who forgot a balance. Concentrate on your vital few, then go find out which cell is quietly paying for it.

The discipline of knowing where value concentrates and who absorbs the cost is in The Unfair Advantage.

Connect on LinkedIn · Book a keynote.