Bad market timing.

Iridium began commercial service in November 1998. The system had been conceived when terrestrial mobile phones were rare and expensive. By launch, cellular coverage had widened, handsets had become cheaper and Iridium was asking customers to carry a costly, heavy phone that worked poorly indoors. The company filed for Chapter 11 protection in August 1999.^[1]

The idea was not imaginary. Global satellite communication serves governments, aviation, maritime operators and remote industry today. The original company was also not killed by the calendar alone. Pricing, handset design, distribution and management mattered. But the timing mechanism is visible: a multi-billion-dollar satellite system took years to build while the substitute market improved faster than the plan.

Founders call many things timing because timing sounds less culpable than demand or execution. This manual uses a harder definition. Timing is causal only when a required external condition is absent or the opening has already closed, the condition was observable at the decision point and the company chose a cost structure that could not wait or differentiate.

That standard assigns responsibility. “The market was early” is not the diagnosis. The diagnosis is that the company leased capacity, hired the organization or promised the rollout before it had a bridge to the present market. The calendar did not sign those commitments.

The opposite failure is lateness. A company sees a validated category and enters through the same product, channel and customer with no asymmetric reason to switch. Here the missing resource is not time to wait. It is an opening. Shipping faster does not recreate one.

Boundary inspection / Timing requires an external prerequisite

“Too early” is not available as a diagnosis merely because the category became large later. The company must identify a prerequisite that customers needed and the startup could not cause on its own: device penetration, network quality, regulatory permission, a standard, an input cost or a habitual behavior. The prerequisite must also explain a specific adoption or economic constraint in the failed model.

If customers could have adopted under existing conditions and a better competitor served them, the diagnosis belongs closer to product-market fit or competition. If the market existed but the startup built too much capacity before proving its own engine, premature scaling may be the primary mode. Timing becomes root when the operating plan makes survival depend on the date of an external event.

The late case uses the same discipline. A validated category is not automatically closed. The company must show that a distribution default, switching cost or incumbent response changed the reachable market before entry. Arriving after press attention is not evidence of lateness. Arriving after neutral access disappears may be.

01 / Forecast-funded capacity

DIAGNOSIS: Warehouses, hardware or staff are built for forecast demand rather than observed utilization.

SIGNAL: The economic model requires future volume to make current fixed cost rational.

INTERVENTION: Add capacity in reversible units tied to utilization, not fundraising.

02 / Missing complement

DIAGNOSIS: The product requires a device, standard, network or behavior the startup does not control.

SIGNAL: Customer activation stalls at a step outside the product.

INTERVENTION: Own the complement, sell through a segment that already has it or pause the market.

03 / Adoption outruns runway

DIAGNOSIS: The company can see readiness improving but cannot survive the slope.

SIGNAL: Each cohort improves while the date of viability moves beyond cash.

INTERVENTION: Build a revenue bridge with the same asset or reduce burn below the readiness curve.

04 / Regulatory countdown

DIAGNOSIS: Revenue assumes permission that has no committed date.

SIGNAL: The plan treats a consultation, pilot or political statement as authorization.

INTERVENTION: Model the company at the current rule; upside from change stays outside the base case.

05 / Cost-curve faith

DIAGNOSIS: Hardware or input economics work only after an expected price decline.

SIGNAL: Gross margin turns positive through a vendor forecast rather than a contracted price.

INTERVENTION: Secure the price, redesign the bill of materials or narrow to buyers who tolerate today’s cost.

06 / Hype-cycle hiring

DIAGNOSIS: Capital availability is interpreted as customer readiness.

SIGNAL: Headcount and market forecasts move together while retained use does not.

INTERVENTION: Separate financing milestones from adoption milestones.

07 / Parity entry

DIAGNOSIS: A late entrant copies the incumbent’s offer and competes for the incumbent’s channel.

SIGNAL: The sales case begins with “like X, but” and the difference is a feature.

INTERVENTION: Find a segment, business model or distribution route the incumbent cannot serve without self-harm.

08 / The permanent bridge

DIAGNOSIS: Services or pilots meant to finance waiting consume the team and never convert into the future product.

SIGNAL: Bridge revenue rises while reusable product evidence remains flat.

INTERVENTION: Give the bridge a margin floor, capacity cap and explicit conversion mechanism.

Review monthly and after every financing decision. A press release does not change a dependency state. The action trigger is the latest survivable date: once the company cannot reach the next evidence event with a downside buffer, it must reduce burn, change the wedge or exit.

Readiness evidence / Forecasts do not move the date

The dependency register accepts production evidence, not category enthusiasm. A signed regulation, installed hardware base, contracted input price or repeated behavior in the target segment can move a readiness estimate. A funding round, conference demo, market forecast or partnership announcement cannot.

Every dependency receives a controller. If the controller is “the market,” the entry is too vague. Name the regulator, platform, supplier, network operator, device manufacturer or customer behavior that controls the milestone. Then name what the startup can observe without asking that party for an optimistic estimate.

Dependencies are combined conservatively. When adoption requires three external events, the base case uses the slowest credible sequence and includes the cost of waiting between them. Multiplying optimistic probabilities produces a financing story; sequencing observed milestones produces a timing plan.

CF-01 / WEBVAN — The warehouse arrived before the volume

• Webvan launched grocery delivery in 1999.
• It built highly automated, purpose-built distribution centers.
• Contemporaneous analysis described demand as weaker than expected.
• The company shut down in 2001 after aggressive expansion.^[3]

Online grocery later became a large category, especially through models that used existing store inventory or denser local infrastructure. That does not vindicate Webvan’s plan. Webvan committed automated capacity that required high, stable utilization before customer behavior had produced it. The future market was attached to the wrong present cost structure.

TRANSFERABLE LESSON: Being early is survivable when the asset grows with demand. Fixed capacity converts timing uncertainty into insolvency.

CF-02 / IRIDIUM — The substitute improved during the build

• Commercial service began in November 1998.
• The network required 66 satellites and billions in capital.
• Handsets were expensive, bulky and weak indoors.
• The company filed for bankruptcy nine months after commercial launch.^[1]

Iridium’s build cycle froze an old comparison: global satellite coverage versus limited cellular service. By launch, the relevant comparison had changed. For most customers, cheap terrestrial coverage beat universal reach with a worse device. The surviving satellite business later concentrated on users for whom global reach was not optional.

TRANSFERABLE LESSON: A long build must re-underwrite the substitute market at every irreversible capital gate.

CF-03 / VREAL — A product downstream of headset adoption

• Vreal built a VR broadcasting and spectator platform.
• It raised about $15 million.
• The company said the consumer VR market had not developed as quickly as expected.
• It closed in 2019 after funding options ended.^[5]

Vreal depended on creators, viewers and a sufficient installed base of capable headsets. The company could improve its software but could not directly create all three. A flat-screen wedge or enterprise use might have changed the exposure; the public record does not establish whether either could support the company.

TRANSFERABLE LESSON: A startup downstream of platform adoption needs a current market that pays it to wait.

The survivor check prevents romanticism. Peapod and later FreshDirect used different operating footprints; the reorganized Iridium served narrower, high-urgency users; many VR studios survived through non-consumer work or flat-screen distribution. The idea did not need a later birth. It needed a present-tense wedge and a cost structure sized to it.

Survivor check / A correct future can support a wrong company

Iridium's later usefulness does not validate the original consumer positioning and capital structure. Online grocery's later growth does not validate building Webvan's footprint ahead of density. Consumer virtual reality can continue improving without proving that Vreal had enough present users and runway to wait. These are cases where category truth and company truth separate.

The survivor comparison therefore holds the future outcome constant and changes the bridge. Which companies entered through a narrow professional segment, leased rather than built capacity, used an existing platform, or waited for distribution to mature? Which commitments remained reversible until utilization appeared? The answer identifies the mechanism that allowed time to pass without making the company insolvent.

This comparison also limits hindsight. Later success may depend on cheaper infrastructure, new devices, a regulatory change or an incumbent's distribution. Those conditions belong in the record. Saying “they were right, just early” erases the operating choices that made early exposure terminal.

The board should be able to state both propositions at once: the destination may be correct, and this company may still be unable to finance the distance. Strategy begins when neither sentence is treated as a contradiction.

Before an irreversible commitment, write:

1. External unlock: the exact condition required.
2. Current baseline: measured today.
3. Control: what the company can and cannot cause.
4. Adoption slope: three cases with provenance, not one forecast.
5. Substitute movement: how alternatives improve during the wait.
6. Irreversible commitments: leases, tooling, inventory, regulation, senior hires.
7. Present wedge: customer who buys at today’s conditions.
8. Latest survivable date: including shutdown or sale time.
9. Re-underwriting cadence: the gate at which the original thesis can be killed.

Operating the memo / Finance the present tense

The timing memo starts with revenue and usage available under current conditions. Future unlocks appear in a separate column and contribute zero to the base case until production evidence crosses the agreed threshold. This prevents a forecast from silently paying today's payroll.

Operations maps every irreversible commitment to utilization: facilities, hardware orders, exclusive contracts and specialized headcount. Capacity is released in units small enough that a missed milestone does not end the company. Where reversibility is impossible, the board treats the commitment as a financing decision and underwrites the full downside.

The memo is refreshed at each capital gate because substitutes keep moving during long builds. A company waiting for its complement must also track improvements in the alternative customers use today. Readiness can rise while the startup's relative value falls. The company survives only if its present wedge or cost structure buys enough time to observe both curves.

Intervention route / maximum 30 days

1. CEO names the missed dependency in one sentence.
2. Finance calculates the latest survivable decision date.
3. Product separates work for current buyers from work for the future market.
4. Sales tests one segment that already possesses the missing complement.
5. Operations halts irreversible capacity additions.
6. Board approves wedge, asset sale, strategic partnership or controlled close.
7. The dependency reopens only with production evidence.

Normal repair works before the company commits capacity that cannot be unwound inside its runway.

1. Cancel forecast-funded leases, inventory and expansion where contractually possible.
2. Renegotiate fixed commitments into usage-based ones.
3. Price the current wedge for today’s costs.
4. Cap bridge work that does not create reusable product.
5. Track substitute improvement alongside market readiness.
6. Reserve cash and time for customer transition.
7. Set a no-fundraising plan before the next dependency review.
8. If no present wedge exists, sell the asset or shut down while records and team remain intact.

Stop rule / The calendar does not accept conviction

The company concentrates, sells or closes when the latest survivable decision date arrives without the required production evidence. The date includes severance, customer transition and contractual obligations; it is earlier than the bank balance reaching zero. A founder may not extend it with a new forecast from the same dependency owner.

One extension is allowed when a current buyer finances the wait at viable economics and the future product reuses the same core asset. Bridge work that consumes all product capacity or requires a second custom organization does not qualify. It is another business masking the original exposure.

The recorded outcome is wedge, wait, asset sale, strategic transfer or controlled close. “Market education” is not an outcome unless customers pay enough to fund it. The purpose of the stop rule is not to abandon a future. It is to prevent an unowned date from consuming every remaining way to reach it.

Timing is not luck with a sophisticated name. The market clock is external, but exposure to it is designed inside the company.

An early startup fails when it builds the organization for the market forecast and asks the current market to carry it. A late startup fails when it enters through parity and assumes a validated category remains an open category. Both mistakes convert time into a claim the company cannot enforce.

The counterintuitive response to being early is not always patience. It is cheapness, narrowness and a buyer who exists now. The response to being late is not haste. It is asymmetry.