Broken unit economics.

In August 2017, MoviePass cut the price of a subscription that could reimburse one cinema ticket every day to $9.95 a month. The offer produced one million subscribers in four months and more than 1.5 million by January 2018. It also bound revenue to a fixed monthly ceiling while an active member could trigger a new ticket expense every day.^[1]

By the first half of 2018, the parent company reported that MoviePass had a significant monthly cash deficit and expected it to increase as subscribers grew. A later regulatory complaint records an average cash deficit of $21.7 million a month disclosed that May. The hoped-for data and ancillary revenue had not become the engine described publicly.^[2][3]

This is the seduction of broken unit economics. Demand appears real. Revenue grows. The product may be loved. The company can therefore narrate the loss as temporary scale investment. But scale helps only when the next unit contributes something toward fixed cost after its own acquisition, fulfillment, support, refunds, incentives and other usage-driven expense.

The root cause is not simply that customer acquisition cost exceeds lifetime value. Both numbers are estimates, and both are often constructed to flatter the plan. The deeper failure is choosing the wrong economic unit and excluding costs or behaviors that make it unprofitable. A marketplace reports gross transaction value while ignoring subsidies. A delivery company measures the parcel but omits pickup labor. A subscription company averages light and heavy users even though heavy use creates the bill.

Broken unit economics occur when leadership scales an incomplete contribution model until each additional unit consumes more cash than it can return inside the available financing horizon. The proximate cause may be a cash crisis or failed round. The root cause is that growth was authorized without an economic unit that reconciled to cash.

Boundary inspection / Price is not the whole equation

This manual covers a structurally unsustainable relationship between customer value and the full cost to acquire and serve it. It does not cover a product that creates value but never asks a payer to fund it; that is monetization failure. It does not cover a viable engine buried by premature headcount or expansion; that is scaling or cash allocation failure.

Losses alone do not prove broken units. A company may rationally spend ahead on a fixed platform, research or a distribution asset while every mature cohort produces positive contribution. The diagnostic question is counterfactual: after removing fixed investment, does an additional comparable unit return cash, and how long does recovery take?

01 / Revenue LTV

DIAGNOSIS: Lifetime value is calculated from revenue rather than gross or contribution profit.

LOAD TEST: Replace revenue with cash collected minus refunds, fulfillment, support, payment fees, incentives and expected service cost.

INTERVENTION: Publish both realized contribution and forecast lifetime contribution by acquisition cohort.

02 / Blended CAC

DIAGNOSIS: Cheap organic customers hide the marginal cost of paid growth.

LOAD TEST: Separate acquisition cost by channel, segment, geography and month; include sales compensation, creative, discounts and agency cost.

INTERVENTION: Authorize spend against marginal cohort CAC, not the company-wide historical average.

03 / The average-user subsidy

DIAGNOSIS: Light users subsidize customers whose usage creates the product's largest cost.

LOAD TEST: Plot contribution by usage decile. Compare the top decile with the median instead of reporting one average.

INTERVENTION: Reprice, meter, cap or redesign the cost-generating behavior before acquiring more of it.

04 / Fulfillment amnesia

DIAGNOSIS: The unit ends at checkout while delivery, returns, damage, fraud and service occur later.

LOAD TEST: Reconcile order contribution after the return and support windows close.

INTERVENTION: Give operations joint ownership of the unit ledger with finance and growth.

05 / Density assumed

DIAGNOSIS: Future route, marketplace or network density is credited to present units without a demonstrated path.

LOAD TEST: Calculate contribution at actual density and at the next reachable density, including the cash required to bridge between them.

INTERVENTION: Prove one bounded zone before using theoretical network efficiency in the plan.

06 / Contracted payback

DIAGNOSIS: The model uses total contract value while cash arrives late or churn can occur early.

LOAD TEST: Compare CAC with cumulative cash contribution at 3, 6, 12 and 18 months.

INTERVENTION: Tie growth gates to cash payback inside the runway, not signed value outside it.

07 / Fixed cost disguised as temporary

DIAGNOSIS: Labor or infrastructure that rises stepwise with volume is labelled fixed.

LOAD TEST: Identify every headcount, kitchen, warehouse, GPU or support step required for the next two volume bands.

INTERVENTION: Model a step-cost curve and require contribution to fund the next step.

08 / Optional revenue completes the model

DIAGNOSIS: Advertising, data, financial services or supplier fees make the forecast work before buyers have committed.

LOAD TEST: Remove every revenue stream without a contract, paid test or repeated transaction.

INTERVENTION: Underwrite the current unit on current revenue; treat new revenue as a separately funded hypothesis.

Choose the smallest repeatable object whose growth drives both revenue and cost: a customer, order, seat, parcel, ride, policy, inference or geographic route. Do not choose whichever denominator makes the deck clean.

Review weekly while changing price, channel or fulfillment and monthly once stable. The action trigger is not a universal LTV:CAC ratio. It is a contradiction: when a cohort forecast as contribution-positive fails to recover its acquisition cost inside the agreed horizon, pause the source of volume until the variance is explained.

The ledger needs three views. The realized view stops at observed cash. The forecast view extends retention and cost assumptions but labels them. The stress view worsens the two least certain assumptions together. A forecast that works only when churn, support and CAC all improve is not a base case. It is a bundle of miracles.

Keep time visible inside every view. Annual revenue collected in advance can make a cohort look self-funding even when the service obligation runs for twelve months. A marketplace may receive customer cash before it pays suppliers. A hardware company may book a deposit before production, returns and warranty claims arrive. Record the cash date and the obligation date separately, then reserve the portion that has not yet been earned. Otherwise working-capital timing will impersonate contribution.

Finally, split the next cohort by acquisition source, product promise and service intensity. A single aggregate can average a sound organic segment with a subsidized paid segment, or low-touch customers with accounts that require repeated human rescue. The purpose is not maximal segmentation. It is to isolate the decision the company can actually stop, price or redesign.

Reconciliation protocol / The unit must reach the bank

Finance begins with cash collected, not booked revenue. Product supplies usage by decile. Operations supplies actual fulfillment and service cost. Growth supplies fully loaded acquisition spend. The four records must resolve to the same cohort.

This catches a common accounting comfort: company gross margin can improve while the newly acquired cohort deteriorates. Legacy customers, annual prepayments or supplier credits may conceal the change. Marginal economics decide whether the next growth dollar is safe.

CF-01 / MOVIEPASS — A fixed price purchased a variable liability

• The August 2017 offer charged $9.95 a month for access to as many as one movie ticket a day.
• Subscribers passed one million in four months.
• The parent reported a significant and rising monthly cash deficit in 2018.
• The plan depended on non-subscription revenue that had not matured.^[1][2][3]

MoviePass did not merely underprice a subscription. It made the most engaged behavior economically dangerous. The member who used the product most could generate repeated full-price ticket reimbursements while subscription revenue stopped at $9.95. Growth increased both subscriber cash and a larger contingent ticket liability.

The company described future value from data, marketing and film-related revenue. Those were not illegitimate ideas. They were separate businesses expected to rescue a unit already underwater. When the service later restricted movie access and changed plans to reduce the deficit, subscriber losses followed.^[1]

TRANSFERABLE LESSON: A future revenue stream cannot be counted as the margin of a present unit until a payer and repeatable transaction exist.

CF-02 / SHYP — The pickup promise omitted the pickup

• Shyp launched with a simple consumer price for collecting, packing and shipping an item.
• It expanded geography and headcount during the on-demand funding cycle.
• CEO Kevin Gibbon later wrote that the company expanded too aggressively before product-market fit.
• Shyp narrowed toward business customers, then closed in March 2018.^[4][5]

The attractive unit was “a shipment.” The costly operation was a pickup, packaging decision, materials, warehouse handoff, carrier purchase, exception handling and support. A flat consumer fee made the experience legible by hiding the variability that operations had to absorb.

The lesson is not that pickup logistics can never work. Shyp found stronger economics in business customers with repeated volume and more predictable workflows. But the correction arrived after expansion had made the consumer model expensive to unwind.

TRANSFERABLE LESSON: Define the unit around the action that creates cost, not the noun the customer buys.

CF-03 / SPRIG — Demand survived; contribution did not

• Sprig prepared meals in its own kitchens and delivered them through its own operation.
• It raised nearly $57 million.
• The company shut in May 2017.
• CEO Gagan Biyani said demand had been high but owning production through delivery at scale was complex.^[6][7]

High demand did not answer whether a meal funded ingredients, waste, kitchen labor, packaging, dispatch, driver time, failed delivery and acquisition. Vertical integration gave Sprig control of experience and ownership of nearly every variable cost.

The public record does not isolate one cohort ledger, so it cannot prove that every meal was permanently negative. It does show the diagnostic boundary: product affection and order volume can coexist with an operation that does not produce sustainable contribution.

TRANSFERABLE LESSON: Operational control creates value only when the price and density pay for the control.

The three cases isolate different unit errors. MoviePass hid usage cost inside a subscription average. Shyp defined the unit too late in the workflow. Sprig owned a chain of costs that demand alone could not retire. None failed because a spreadsheet ratio crossed a universal threshold. Each let the customer promise outrun the cash mechanics of delivering it.

Before any material growth commitment, the CEO, finance owner and operating owner sign one page:

1. Unit definition: the object whose growth creates revenue and cost.
2. Cohort: segment, channel, geography and start month.
3. Realized contribution: cash collected less every volume-linked cost.
4. Service tail: the remaining obligations after the first transaction.
5. Payback clock: when cumulative cash contribution recovers acquisition.
6. Step costs: capacity added at the next two volume bands.
7. Unproved upside: revenue or efficiency excluded from the base case.
8. Growth gate: spend or volume released only when evidence holds.
9. Stop owner: executive empowered to pause acquisition or fulfillment.

The warrant is not an annual plan. It expires when price, channel, customer, product promise or fulfillment method changes materially. Those changes create a new unit and require a new cohort.

Cohort release / Growth is earned in tranches

Release a bounded amount of acquisition or capacity. Wait long enough to observe the cost tail and early retention. Reconcile the cohort to cash. Then continue, revise or stop.

The board receives the realized, forecast and stress views together. If management presents only blended company economics, the board asks for the newest marginal cohort. If the unit cannot be observed before the company needs to commit the next capacity step, the batch is too large.

Intervention route / maximum 60 days

1. The CEO freezes new discretionary acquisition and capacity commitments for the disputed unit.
2. Finance defines one cohort and reconciles its cash revenue.
3. Operations assigns every usage-driven and step cost to that cohort.
4. Growth rebuilds CAC with all channel, labor, discount and creative cost.
5. Product reports retention and usage by decile rather than one average.
6. Leadership runs one bounded lever: price, package, segment, channel or fulfillment.
7. A 30- and 60-day review chooses reopen, revise once, shrink to a viable segment or stop.
8. Only reconciled cohort evidence reopens the growth gate.

One lever matters. Simultaneously raising price, cutting service, changing segment and switching channels may improve the cash line, but it destroys the diagnosis. The company no longer knows which mechanism worked.

Normal repair remains possible when the company can pause marginal growth, observe a full enough cohort and still fund a pricing or operating test. Once future service obligations exceed available cash, containment comes first.

1. Stop incentives and channels attached to contribution-negative cohorts.
2. Protect cash required for refunds, credits, payroll and contracted service.
3. Reconcile deferred revenue and future usage obligations.
4. Rank segments by realized contribution, not total revenue.
5. Preserve the smallest segment with positive or credibly repairable units.
6. Reprice or narrow the promise for new customers before changing existing contracts.
7. Renegotiate suppliers and capacity only against an observed volume band.
8. Give customers a clear migration, limitation or closure path.
9. Close the unit when one permitted revision cannot produce cash payback inside the remaining runway.

Containment rule / Do not finance usage you cannot honor

Subscription cash can arrive before service cost. Marketplace balances can belong to sellers. Insurance-like promises can create claims after revenue is booked. These balances are not available growth capital merely because they are in the bank.

The repair plan therefore separates unrestricted cash from customer and counterparty obligations. Legal and accounting advice may be required before changing paid plans, holding marketplace funds or preferring one creditor class. Economic diagnosis does not override those duties.

Broken unit economics are not a synonym for spending money before profit. They are a growth contract the company cannot afford to fulfill.

MoviePass made heavy use generate a large uncapped cost against a capped price. Shyp's simple shipment price hid the labor and variability before the carrier. Sprig owned production and delivery without proving that the meal could fund the chain. In each case, activity was visible before contribution.

The liberating implication is that the whole company may not be broken. The false unit may be a channel, segment, geography, usage band or promise. A precise ledger can shrink the problem from “our model does not work” to “this customer acquired this way cannot fund this service level.” That is an actionable sentence.

Verdict equation: FULLY LOADED CAC + COST TO SERVE + SERVICE TAIL > REALIZED CONTRIBUTION → GROWTH COMPOUNDS CASH LOSS