There is a category of restaurant industry writing that gets produced with the confidence of pricing analysis and the frame of a Road 1 instrument, and it looks like this: the writer identifies a loyalty program mechanic that extracts more margin from the operator’s most loyal customers, names the extraction with technical precision, and closes by recommending the extraction as best practice.
The recent case study of this pattern is a piece prosecuting Crumbl’s 2024 loyalty program redesign as a “regressive redemption ladder.” The author’s analytical work is clean. He shows precisely how Crumbl replaced a straightforward 10% rebate with a tiered points-based program whose effective rebate is 3-6% depending on tier and redemption path, and whose most-loyal customers are systematically punished for the exact behavior the program is telling them to run. The piece extends the same analysis to Dunkin’, Starbucks, and Marriott Bonvoy, and identifies the same pattern across all four operators. The evidence is solid. The pattern-work is disciplined. The conclusion is unambiguous.
Then the author writes this: “moderately complex price structures are far more effective for the business than simple ones.” And: “this program revamp is entirely consistent with my assertion about pricing best practice.”
That is the moment the piece stops being analysis and becomes propagation. The mechanism the piece just prosecuted as extractive gets renamed as effective. The failure mode gets prescribed as best practice. And every operator reading the piece will treat it as instruction. Add tiers. Add product-specific redemption. Add expiration. Add opacity. Give yourself more pricing levers.
That instruction is a hack. This piece prosecutes the hack, names the category error underneath it, and shows what a Road 2 loyalty program actually is.
The Loyalty Program Category Error
Every argument the pricing chair makes about loyalty programs rests on a single assumption: that a loyalty program is a pricing instrument. The academic literature treats it as one. The consultant class treats it as one. Every operator who has hired a “loyalty program strategist” in the last decade has been sold one. And the entire optimization apparatus — earning rates, redemption ratios, tier thresholds, expiration policies, product eligibility, dynamic pricing extensions — flows from that assumption.
That assumption is the category error. It is the assumption that produces the hack.
A loyalty program on Road 1 is a pricing instrument. Its purpose is to manage the effective price a segment of customers pays across a sequence of purchases. Its levers are earning rates, redemption ratios, tier thresholds, expiration policies. Its optimization objective is to maximize captured value per redemption while producing enough perceived generosity to sustain the program’s participation rate. The customer is a segment. The program is a discrimination engine. The relational language (“valued customer,” “loyalty,” “rewards”) is marketing collateral wrapped around a pricing mechanism.
A loyalty program on Road 2 is not a pricing instrument. It is relational infrastructure. Its purpose is to make the operation’s Guest Contract legible to the Guest, and to surface which Guests are worth further investment. Its levers are the operation’s Voice of the Guest listening system, the operation’s product decisions, the operation’s cast development, and the operation’s investment in Guests who have chosen to invest in the operation. Its optimization objective is to strengthen the relational contract between the operation and its Guests over time. The Guest is not a segment. The program is not a discrimination engine. And the language is not collateral — it is the operating truth of what the program is.
Every loyalty program built in QSR in the last twenty years has been built on Road 1. Every academic paper written about loyalty programs treats them as Road 1 instruments. Every consultant selling loyalty program design is selling Road 1 mechanics. The category error is total, and the industry no longer has the vocabulary to name it.
That is the arbitrage. The industry has forgotten what a loyalty program was supposed to be.
The Regressive Redemption Ladder Is A Contract Violation, Not A Design Tradeoff
Crumbl’s revamped program has a specific and diagnosable failure at the Guest experience level. The Guest who accumulates Crumbs — the exact behavior the program is telling the Guest to run — is punished for accumulating them. A single cookie costs 500 Crumbs. A 12-pack costs 6,000 Crumbs. On the cash menu, the 12-pack carries a 20% quantity discount. On the rewards menu, the quantity discount has been amputated. The Guest who runs the program’s advertised behavior receives less value per Crumb than the Guest who redeems impulsively.
The pricing-chair analysis calls this a “regressive redemption ladder” and treats it as a design tradeoff — the operator gets pricing flexibility, the Guest gets less value, both parties made a trade. That framing is a Road 1 read. It presumes the loyalty program is a pricing negotiation between operator and customer, and it evaluates the negotiation on whether the operator extracted appropriately.
The Road 2 read is that this is not a negotiation. It is a contract violation.
Every operation has a Guest Contract. The Guest Contract is not a legal document. It is the sum of every promise the operation has made to the Guest — explicit and implicit, spoken and implied by the operation’s design — and it is what the Guest is signing every time they walk in, order, and pay. A loyalty program is one of the most explicit promise-making instruments the operation has. It says to the Guest: “invest in this operation, and this operation will invest in you. Save your rewards, and you will earn something worth saving for.”
Crumbl’s revamped program makes that promise and breaks it in the same document. It says to the Guest “save your Crumbs” and then structures the redemption menu so that saving Crumbs punishes the Guest. It says to the Guest “our best customers get our best deals” and then makes the best customers pay the highest effective per-cookie rate on the reward they were told to save for. It says to the Guest “you are earning ten times as many points” and then hides a per-Crumb devaluation that is many times worse than the earning-rate improvement.
That is not a design tradeoff. That is the operation making a promise to the Guest and structuring the mechanics of that promise so that the Guest cannot collect on it. The operation has broken its Guest Contract with the exact Guests it named as most valuable. And it has done so in a way that most Guests cannot detect without algebra.
Dunkin’ Rewards is the same violation. The 2022 redesign lowered the entry threshold for a reward (150 points instead of 200) and simultaneously raised the beverage reward threshold from what used to be $40 of spending to $60-$95 of spending. The Guest who used the old program to earn a free beverage every seven visits now needs 12 to 18 visits for the same beverage. The Guest whose relationship with Dunkin’ was built on that beverage-earning cadence had that cadence stolen. The reciprocal promise the old program made — you buy coffee, we reward you with coffee — has been dismantled. The mechanism the pricing chair calls “recognizing our best customers” is the mechanism that punishes the operation’s most loyal beverage-purchasing cohort the hardest.
Starbucks is the same violation with a decade of head start. The 2023 redesign doubled the Stars required for the two categories that produce the majority of Starbucks’ daily-visit Guests — brewed coffee, tea, small bakery. Those are the Guests whose daily-visit habit is the operation’s most defensible asset. Doubling the Stars required for the reward they earn most frequently is a targeted price increase on the exact behavior the operation depends on. The 2026 tiered redesign compounds the violation by making the newest Guests earn Stars slowest and the highest-tier Guests earn fastest — a mechanic that is only defensible if you accept the pricing-chair framing that loyalty programs exist to price-discriminate. If you accept the Guest Contract framing, the mechanic is prosecuting the operation’s newest Guests for the crime of being new. That is not a reward program. That is a penalty program with reward language painted over it.
Marriott Bonvoy’s dynamic redemption is the endpoint. The Guest can no longer determine what an accumulated point is worth. The exchange rate fluctuates. The value floats with the operator’s real-time margin optimization. The Guest is not saving toward a known reward. The Guest is holding a currency the operator can devalue at will. The pricing chair calls this “flexibility.” The Guest Contract framing calls this the total surrender of the relational contract to the transactional instrument.
Every one of these programs is doing the same thing at the architecture level. Every one of these programs has abandoned the relational contract and become a pricing engine wearing a loyalty program’s costume. And every one of them is presented to operators as best practice.
The Industry Infrastructure That Keeps Producing The Hack
The Chipotle-of-X piece named six layers of industry infrastructure that keep the Framework Arbitrage pattern alive. Loyalty program arbitrage has its own infrastructure, partially overlapping and partially distinct. Six layers keep it alive.
The academic pricing literature. Marketing academics writing about loyalty programs treat them as pricing instruments by disciplinary convention. The frame is inherited. The Journal of Retailing and Consumer Services, the Journal of Consumer Research, the Journal of Marketing Research — every article on loyalty programs in the last twenty years reads them as pricing instruments and evaluates them on pricing-outcome metrics. The relational-contract reading does not appear in the literature because the literature’s frame does not permit it. The frame is the arbitrage propagating through the academic pipeline.
The pricing consultancy class. Firms and independent consultants who sell pricing strategy to QSR operators sell loyalty program redesigns as pricing optimization projects. The engagement structure is: audit the current program, identify extraction opportunities, recommend a tiered restructure with product-specific redemption, implement, measure margin lift. Every part of that engagement runs on Road 1 physics. The relational contract is not a variable in the analysis.
The loyalty-tech platforms. Punchh, Paytronix, Thanx, and every other loyalty-tech vendor sells a product whose feature set is a pricing-instrument feature set. Tiers. Points. Redemption menus. Expiration. Dynamic promotions. The platform architecture presumes the program is a pricing instrument and cannot support a program that is not. Operators buying these platforms are locked into Road 1 mechanics at the technology layer before they have made a single design decision.
The trade press coverage. Restaurant industry trade coverage of loyalty program redesigns evaluates them on pricing-outcome metrics — points earned per dollar, redemption ratios, engagement rates, incremental revenue per member. The relational-contract lens does not appear. Every operator reading trade coverage of a competitor’s loyalty program is being trained to evaluate their own program on the same terms.
The private-equity thesis-copy. When PE firms buy QSR concepts, one of the first initiatives is “monetize the loyalty program” — which in practice means restructure it toward higher extraction. The playbook is inherited from firm to firm and from operating-partner to operating-partner. The Road 2 loyalty program is not on the playbook because the playbook was written from the Road 1 chair.
The business-school case method. Marketing programs at business schools teach loyalty programs through case studies of Sephora, Starbucks, Marriott, and Amazon Prime — all Road 1 instruments in various stages of sophistication. Students graduate believing that a loyalty program is a pricing instrument because the only examples they have ever studied are pricing instruments. When those students become brand managers and consultants and PE operating partners, they build more Road 1 loyalty programs.
Six layers. Each reinforcing the others. Each producing operators who cannot see the category error because the vocabulary to name it has been removed from every institution operators learn from.
The Rationalizations Operators Deploy In Retreat
When an operator’s revamped loyalty program produces the failure modes named above — the regressive ladder, the Guest Contract violation, the extraction from the operation’s most loyal cohort — the operator rarely dismantles the program. The operator deploys a rationalization. The rationalizations are predictable and enumerable. Eight of them recur.
“The program is more generous now.” This is the earning-rate defense. The operator points to the 10X or 13X earning-rate improvement and treats it as sufficient evidence of generosity. The redemption-side devaluation is not addressed because the earning-rate improvement is more visible.
“Our best customers get the best deals.” This is the tier defense. The operator points to the Pink or Reserve or All-Star tier’s marginally-better economics and treats it as sufficient evidence of relational investment. The fact that the tier’s marginal advantage is a small fraction of the extraction happening across the whole program is not addressed.
“Customers wanted more flexibility.” This is the choice defense. The operator points to the expanded product-specific redemption menu and treats it as sufficient evidence of Guest empowerment. The fact that the “flexibility” is a decoy for a decoupled price schedule the Guest cannot easily evaluate is not addressed.
“The old program was too expensive to operate.” This is the cost defense. The operator points to the redemption expense of the old program and treats it as sufficient evidence that the new program is necessary. The fact that the operating cost of the old program was the operation’s investment in its relational contract — not an expense but a capital allocation — is not addressed.
“Complexity is best practice.” This is the borrowed-authority defense. The operator points to the academic literature, the consultant recommendations, the trade press coverage, and treats them as sufficient evidence that complexity produces better outcomes. The fact that the entire body of authority was built on the Road 1 frame is not addressed.
“Engagement is up.” This is the metric defense. The operator points to program signup rates, in-app notifications delivered, points-earned notifications viewed, and treats them as sufficient evidence of program health. The fact that engagement metrics measure attention to the extraction mechanism, not satisfaction with the relational contract, is not addressed.
“The math is complicated for a reason.” This is the sophistication defense. The operator points to the multi-dimensional structure of the program — tiers, product-specific redemption, expiration, dynamic pricing — and treats it as sufficient evidence that the program is professionally designed. The fact that the complication is what produces the Guest’s inability to evaluate the program is not addressed.
“If it were really bad, customers would leave.” This is the retention defense. The operator points to the program’s participation rate and treats it as sufficient evidence that the Guest is satisfied. The fact that Guest churn from a loyalty program is a lagging indicator that arrives long after the relational contract has been broken — often years after — is not addressed.
Each rationalization is a Road 1 defense against a Road 2 read. None of them address the underlying frame. All of them keep the operator on Road 1.
What A Road 2 Loyalty Program Actually Is
A Road 2 loyalty program is designed from the Guest Contract, not from the pricing chair. Six architecture principles distinguish it.
Principle 1 — The program’s value proposition is knowable in a single sentence. The Guest can state, in one sentence, what they are earning and what it will be worth. “For every dollar I spend, I get ten cents back to spend on anything on the menu.” That is a Road 2 promise. It is legible. It is unconditional. It cannot be devalued without the Guest immediately detecting the devaluation. Every layer of complexity the pricing chair calls “sophistication” is a layer of illegibility to the Guest. Illegibility is the arbitrage. Legibility is the contract.
Principle 2 — The reward mechanic reinforces the behavior it names. If the program tells the Guest to save toward a bigger reward, the bigger reward must be objectively better than the smaller reward on a per-unit basis. If the program tells the Guest to concentrate visits at the operation, the reward structure must reward concentration and not punish it. If the program tells the Guest that longer relationships are valued, the reward structure must value longer relationships and not amputate them at annual reset. The mechanic must do what the mechanic says it does.
Principle 3 — The program’s economics are stable across time. The Guest who joins the program in January and the Guest who joins in December face the same reward structure, the same earning rate, and the same redemption values. The Guest whose reward balance is a year old can redeem that balance for the same value as the Guest whose reward balance is a month old. Expiration policies, dynamic pricing, tier resets — every mechanic that introduces temporal instability into the program’s economics is a Road 1 mechanic. Road 2 programs do not require temporal instability to be commercially viable. They require good operating architecture.
Principle 4 — The program’s tiering, if tiering exists, tracks the operation’s actual investment in the Guest, not the Guest’s spending threshold. A Road 2 tier is a designation of Guests the operation has chosen to invest in further — Guests who receive named cast attention, early access to new products the operation is developing, direct input into the operation’s Voice of the Guest system. A Road 2 tier is not a spending threshold that triggers a slightly-better earning rate. If the tier only produces marginal pricing improvements, it is a Road 1 pricing tier with a Road 2 name attached.
Principle 5 — The program produces intelligence the operation acts on. A Road 2 loyalty program is a Voice of the Guest listening system. It surfaces which Guests return, which products drive return, which visit patterns predict long-term relationship, and which Guests are signaling capacity for further investment. That intelligence is not used to price-discriminate. It is used to invest in the Guests the operation should be investing in — through product decisions, cast attention, and hospitality architecture. A program that produces intelligence and only uses that intelligence to optimize price is a Road 1 program.
Principle 6 — The program’s design is coherent with the operation’s design. A Road 2 loyalty program is not a bolted-on module. It is one instrument in an operation-wide architecture. The program’s reward structure reflects the operation’s product priorities. The program’s cast interactions reflect the operation’s hospitality standards. The program’s Guest-facing communications reflect the operation’s brand voice. The program is a component of the operation’s coherence, not a separate optimization system running in parallel. When the loyalty program is designed by a pricing consultant on a separate engagement from the operation’s other decisions, it produces incoherence. Incoherence is the tell.
Six principles. Each one is a rejection of one of the Road 1 defaults. Together they name what a loyalty program looks like when it is designed from the Guest Contract instead of from the extraction chair.
The Operator Test
Every operator in QSR can run this test against their own loyalty program before the end of this week.
Test One — The Sentence Test. Ask any Guest at the counter, or on the app, to state in one sentence what your loyalty program gives them. If they cannot state it in one sentence, the program has failed the legibility test. If they state it wrong, the program has failed the legibility test. If they state it correctly but with hedges (“I think it’s…”, “I’m pretty sure it depends on…”), the program has failed the legibility test. A Road 2 loyalty program passes this test with the first Guest asked.
Test Two — The Saver Test. Run the math on your own program: does the Guest who saves rewards to accumulate a bigger reward get more value per point than the Guest who redeems small rewards frequently? Or does the Guest who saves get less? If saving produces less value per point, the program is punishing the behavior it advertises. That is a contract violation. Fix it or dismantle the program.
Test Three — The Tenure Test. Compare the effective rebate rate a Guest received under your old program with the effective rebate rate the same Guest is receiving under your new program. If the new program’s effective rebate is lower — even if the earning rate is higher — the program has extracted from the Guest. The Guest may not have noticed yet. The Guest will notice eventually. When they do, the relational contract will be broken retroactively across the entire duration of the relationship, not just the extraction period.
Test Four — The Cast Test. Ask your cast members what your loyalty program is for. If they say “getting people to sign up for the app” or “meeting our targets,” the program is a Road 1 program at the internal-communication level. If they say “recognizing our best Guests” or “investing in the people who invest in us,” the program has at least the language of a Road 2 program. If the language is Road 2 but the mechanic is Road 1, the program is manufacturing cognitive dissonance in the cast every day.
Test Five — The Consultant Test. Ask whoever designed your loyalty program what happens to your program’s economics if you eliminate expiration, eliminate product-specific redemption, and convert every point to a fixed dollar value redeemable against any menu item. If the answer is “the program becomes commercially unviable,” your program is running on Road 1 mechanics that only work because of illegibility. If the answer is “the program becomes more attractive but slightly more expensive to operate,” the program can survive the conversion to Road 2. If the answer is a defensive posture about “industry best practices,” you have identified the arbitrage in your own building.
Test Six — The Dholakia Test. Read a pricing-strategy analysis of your loyalty program. If the analysis prosecutes your program as extractive and simultaneously recommends the extraction as best practice, you have located the exact category error propagating through the industry. The analyst is not wrong about the extraction. The analyst is wrong about it being best practice.
What Changes Tomorrow
The operator running a QSR concept with a redesigned loyalty program has three moves available tomorrow morning, in order of difficulty.
The first move is the audit. Run the six tests above against your own program. Do not hire a consultant for this. Do not schedule a working group. Read the six tests. Answer them honestly on a single page. If your program fails any test, you have work to do.
The second move is the conversation. Talk to the person or firm who designed your program. Ask them the questions the six tests raise. If they respond with pricing-strategy vocabulary — “extraction opportunities,” “tier optimization,” “engagement lift,” “redemption cost management” — they are running on the Road 1 frame and they will resist the audit’s implications. If they respond with relational vocabulary — “Guest contract,” “long-term investment,” “cast alignment,” “operation coherence” — they are already partially on Road 2 and the conversation can go somewhere productive.
The third move is the redesign. A Road 2 loyalty program can be built from a Road 1 program. It requires converting product-specific redemption to fixed-dollar redemption. It requires eliminating expiration. It requires either eliminating tiers or converting tiers from spending thresholds to investment designations. It requires integrating the program’s Voice of the Guest signal into the operation’s actual product and cast decisions. It is more expensive to operate than the Road 1 program because it does not extract from the Guest to fund itself. It is more valuable to the operation than the Road 1 program because it produces the relational infrastructure the operation actually competes on.
None of the three moves happens without the operator first accepting that the loyalty program is not a pricing instrument. That acceptance is the choke point. Every Road 1 loyalty program in QSR exists because the operator has not made that acceptance. And every academic paper, consultant engagement, tech platform, trade press article, PE playbook, and business-school case is arrayed to prevent the operator from making it.
The industry is not going to make the acceptance for you. The industry is invested in the pricing-instrument framing because the industry has built its business models on it. The operator makes the acceptance alone, in the operation, on a Monday morning, with a single page in front of them and the will to look at what they built and name what it is.
That is the move. That is the road back.
Digging Deeper
Positions On The Record
- The Chipotle Of X Is Framework Arbitrage
- The Tool Stack Is Not A Framework
- Two Roads
- The Guest Contract Is The Operation
- Voice Of The Guest Is A System, Not A Survey
Term Definitions From The Knowledge Base
- [Transactional Arbitrage]
- [Framework Arbitrage]
- [The Guest Contract]
- [Two Roads]
- [VoG]
- [Voice Systems]
- [Case Study Reduction]
- [Architectural Coherence]
Sources
- The Regressive Redemption Ladder In Crumbl’s Loyalty Program — Utpal Dholakia, Psychology of Pricing (Substack), August 15, 2026
