The Cascading Miracles Trap

A business that works only if a long chain of hard, sequential bets pays off: each one necessary, none sufficient, and the combined odds quietly stacked against the founder.

The math is not a metaphor; it is the point. Suppose a plan depends on five things going right, and you are genuinely optimistic about each one: every link gets an 80% chance. The chain’s odds are not 80%. They are 0.8 multiplied five times, which is about 33%. Move to a more honest 60% per link and the chain drops to roughly 8%. Founders rarely run this multiplication because they evaluate each bet alone, and each one looks winnable. The trap is not any single improbable step. It is the structure that strings several hard steps together and lets the compounding hide in plain sight.

Symptoms

The trap is hard to see from inside because every individual assumption looks reasonable. It shows up in the shape of the plan, not in any one of its parts. Watch for these together.

• The pitch is a sequence of “and then.” The story only reaches the payoff after a string of dependent steps: “first we sign the manufacturers, and then we build the network, and then consumers switch, and then the ad market follows.” Each clause is a separate bet, and each later one is dead if an earlier one misses.
• Success requires changing several actors’ behavior at once. The model needs customers to adopt a new habit and suppliers to commit and a partner network to form and, often, a regulator to bless it. None of these parties moves first without the others, so the venture has to manufacture all of them in the right order.
• The capital required to test the thesis is enormous. You cannot learn whether the chain holds with a cheap experiment. The early links (the factory, the network, the hardware) have to exist before the later links (adoption, revenue) can even be observed. The first real test of the model is also the one that bets the company.
• Every milestone is a relief rather than a result. The team celebrates clearing each step not because it created value but because it kept the chain alive. Progress is measured in miracles survived, not in a working business getting more obviously viable.

A single ambitious dependency is normal; many worthwhile startups bet on at least one hard thing. The trap is the plan that needs three, four, or five hard things to land in order, with no version that works if any of them slips.

Why It Happens

Cascading Miracles is not a failure of nerve or intelligence. It happens when smart, ambitious founders evaluate a model one link at a time and miss the structural flaw in the chain.

The first cause is that founders evaluate links, not chains. Asked “can we solve the battery problem?” or “can we sign the first manufacturer?”, a capable team answers each question honestly and optimistically, one at a time. What almost no one does at the whiteboard is multiply the answers together. The mind treats a sequence of plausible steps as a plausible plan, when a chain of likely-enough bets is far less likely than any of its parts. The optimism that makes founders found is exactly what blinds them to the product of the probabilities.

The second cause is that the biggest opportunities are often shaped this way. The markets worth a venture’s effort are often the ones no one has unlocked precisely because they require several things to change together. A founder may be right that the prize is huge and right that no incumbent has taken it. The error is concluding that they should be the one to make all the miracles happen. The size of the prize is not evidence that the chain will hold. It’s usually the reason the chain is so long.

The third cause is Knightian uncertainty stacked on itself. Each link in a cascading model is typically a true unknown rather than a measurable risk. The question is not “will this coin land heads” but “will an entire category of customers adopt a behavior they have never had.” You cannot price one such bet honestly, let alone five compounded. So the team substitutes confidence for a probability it cannot actually compute, and confidence multiplies a lot more comfortably than 0.6 does.

The Harm

The company spends years and a fortune to discover what the structure implied at the start. Betting on a long chain of independent miracles is a low-probability strategy, however good each bet looked alone.

The most direct cost is capital, because cascading models are expensive by nature. The early links have to be built before the later ones can be tested, so the venture raises and burns heavily to construct the factory, the network, or the hardware that the whole thesis depends on. When a late link fails (consumers don’t switch, the ad market never forms), the spend on the early links is unrecoverable. There is no graceful retreat, because the early investment only made sense if the later miracles arrived.

The second cost is time, and it falls hardest on the founders and the team. A cascading venture can run for five or seven years clearing miracle after miracle before the chain breaks at a link no one can force. Each cleared step renews everyone’s belief. Those are years the founders cannot get back and cannot easily redeploy, because the skills and assets were specialized to a model that turned out not to work.

The cruelest version is the venture that fails on its last miracle. A team can solve the hard technical problem, sign the manufacturers, and build the network, and still die because the final link, mainstream adoption, simply doesn’t come. The post-mortem reads like a tragedy of inches: “they did everything right and still lost.” But the loss was probabilistic, not accidental. A strategy that needs five things to go right will, most of the time, fail on one of them, and which one fails is close to noise.

The Way Out

The exit is not “be less ambitious.” Some of the most valuable companies ever built were cascading bets that happened to land. The discipline is to see the chain clearly, then either shorten it, sequence it so each link is cheap to test, or take the bet with eyes open rather than by accident.

First, map the chain and multiply. Before committing, list every assumption the model needs to be true and assign each an honest probability. Then multiply them. The exercise is uncomfortable on purpose: a plan that needs five 70% bets is a 17% plan, and seeing that number is the first time most teams reckon with the structure they have signed up for. If the multiplied odds are intolerable, the question is no longer “how do we execute” but “how do we change the model.”

Second, attack the chain’s length, not just its execution. The strongest move is to find a version of the business that delivers value after the first link, not the fifth. Can the technology be sold into an existing market that needs no new behavior, funding the company while the longer bet matures? Can one miracle be bought or partnered away instead of performed? Every link you can remove, de-risk, or defer turns a cascade into something closer to a single bet, and single bets are what startups are built to win.

Third, if the chain is irreducible, fund and sequence it as the long, low-odds bet it is. A few businesses genuinely cannot be unbundled: the value only exists once every link holds. Those can still be worth pursuing, but only by a team and investors who have run the multiplication, accepted the odds, raised enough patient capital to reach the final miracle, and tested the cheapest fatal link first. The fastest way to lose a cascading bet is to spend the capital on the early, buildable links and arrive at the hardest one, usually customer behavior, with the treasury already empty.

How It Plays Out

The original Iridium is the clean public case. Backed by Motorola and launched in the 1990s, it bet on a chain of miracles that each had to hold. The constellation of 66 low-Earth-orbit satellites had to work. Terrestrial cellular networks had to remain sparse long enough for the service to matter. Customers had to carry the handset. Enough travelers had to pay a premium for a phone that worked anywhere on Earth. Motorola and its partners cleared the hardest technical links: the satellites flew. But by the time the service launched in 1998, the ground networks the plan had bet against had spread across the markets customers actually traveled in. The handset was bulky, calls were expensive, and mainstream demand never materialized. The company filed for bankruptcy within a year, having spent roughly $5 billion to discover that the last link in the chain, the customer, would not hold. The miracles that could be engineered were; the one that depended on millions of people changing their behavior was not.

The quieter version plays out in marketplace and platform startups every year. A team sets out to build a two-sided market that needs suppliers to list and buyers to show up and a cold-start chicken-and-egg problem solved and a network effect to kick in before the cash runs out. Each piece is a known startup challenge, and the founders are confident they can solve any one of them. They raise on the size of the eventual prize, build the supply side, then build the demand side, and discover that the network effect they were counting on arrives, if at all, a year after the runway ends. No single link was impossible. The model needed too many of them to land in the right order, on a budget that only covered the early ones.

Related Articles

Sources

• Tom Eisenmann, Why Startups Fail (2021) — the Harvard Business School research that names the Cascading Miracles failure mode among six recurring archetypes and traces how chained, dependent bets sink ventures that solved every hard problem but one.
• Frank Knight, Risk, Uncertainty and Profit (1921) — the distinction between measurable risk and genuine uncertainty that explains why the links in a cascading model cannot be honestly priced or multiplied with confidence.