Anatomy of costly decisions: lessons from six-figure business mistakes (or more)

Series: Decisions that Move the Needle: Article 10 of 10

Business mistakes that make history are rarely stupid decisions. That is the most uncomfortable part of their anatomy: at the moment they were made, almost all of them seemed reasonable, backed by real data, defended by brilliant teams, and supported by the best practices of their time. The fear of making mistakes is not cured with more generic prudence; it is cured by understanding how smart decisions turn into nine-figure losses.

This article dissects four of the most studied mistakes in recent business history, with the documented figures for each case, and extracts the common pattern, because the pattern, not the case, is what translates to an SMB of any size. We close with the statistic that frames everything: the evidence on how most major projects end when planning fallacy is not managed.

The framework: what the evidence says about major projects

Bent Flyvbjerg built the largest database of major projects in existence (more than 16,000 projects in his Oxford archive), and his conclusions are the required starting point. Only 8.5% of projects finish on budget and on time; barely 0.5% meet budget, schedule, and promised benefits (Flyvbjerg, project database). In his book How Big Things Get Done (2023), the summary is even more striking: 92% of megaprojects end over budget, over schedule, or both (Penguin Random House on the book).

Translated: optimistic planning is not the exception in major business decisions: it is the statistical norm. Each of the cases that follow is, on a different scale, a variation of that norm.

Case 1: Kodak: the decision to protect the past

The numbers. Kodak employed 145,300 people at its 1988 peak (Rochester Business Journal, 2017). In 1975, its own engineer Steven Sasson had invented the first digital camera, inside Kodak (Forbes, 2012). In January 2012, the company filed for bankruptcy.

Why it seemed reasonable at the time. The decision not to bet aggressively on digital photography was not born of ignorance: it was born of the arithmetic of the existing business. Film was a business with very high margins and perfect recurrence (every roll consumed generated more consumption); the digital camera cannibalized that business with lower initial margins. Defending current cash flow over an emerging technology was, by the financial analysis standards of the time, the prudent decision.

The transferable lesson. The mistake was not "failing to see" digitization: they saw it so closely that they invented it. It was evaluating the decision with a single question: "what happens to my current business?" instead of the correct question: "what happens if this technology develops without us?" Calculated cannibalism (having your own new products eat your old products) always looks irrational on the quarterly spreadsheet, and it is the only defensible position over a decade. In an SMB, the everyday version of this mistake is rejecting a channel, a model, or a technology because it "breaks what already works," without evaluating the scenario where competitors break it.

Case 2: Blockbuster: the decision not to buy the future cheaply

The numbers. In 2000, Netflix offered to sell itself to Blockbuster for US$50 million; Blockbuster rejected the offer. A decade later, in 2010, Blockbuster went bankrupt, unable to pay a debt on the order of US$1,000 million (CNBC, 2020).

Why it seemed reasonable at the time. Seen from 2000, Netflix was a marginal business: tiny revenue, niche market, a mail-order model unproven at scale. Late fees, which seem absurd today, were a structurally significant fraction of rental profitability. Declining to buy a small and expensive competitor for a market that was "always" rented the same way was a defensible decision with the available data, and in fact, many people defended it.

The transferable lesson. Here the mistake was not one of analysis but of time horizon: Blockbuster evaluated Netflix as it was, not as it could be. In strategic decisions, the current size of a competitor or opportunity is the least informative figure; the model's rate of improvement (its cost/quality curve) is what matters. The SMB version of this mistake: dismissing a channel or a technology because "it is small today," without monitoring its growth rate. The operating rule is simple: whatever grows fast and improves the customer experience never stays small.

Case 3: Quibi: the decision to fund the answer before validating the question

The numbers. Quibi raised US$1,750 million before launching its mobile video platform, launched in April 2020, and announced its shutdown in October 2020, barely six months later, without having managed to attract the projected subscriber base (CNBC, 2020).

Why it seemed reasonable at the time. This is the most instructive case for entrepreneurs because everything else was done right: a world-class team, content with A-list talent, solid technology, abundant capital. The thesis (premium video in short mobile format) had logic and even some backing from trends. What was missing was cheap, early validation of the central hypothesis (would people pay for this?) before committing capital to mass production.

The transferable lesson. Capital does not compensate for a lack of validation; it amplifies its cost. The evidence we saw in the article on pivoting or persevering applies directly: 70% of startups scale prematurely in some dimension, and that premature scaling is the primary cause of failure in the Startup Genome database (Startup Genome). Quibi is the extreme case: validating after spending US$1,750 million (CNBC, 2020) is validating a cemetery. At any scale, the correct sequence is the same: hypothesis → cheap test → decision → and only then, scaled investment.

Case 4: the cross-cutting pattern: structural planning optimism

The fourth case is not a company: it is the statistic that runs across all of them. Returning to Flyvbjerg's database: less than 1% of major projects meet budget, schedule, and promised benefits (Flyvbjerg). And in the startup failure research we already cited, CB Insights' analysis of 431 post-mortems shows that "running out of cash" appears in 70% of cases as the final cause, the accounting symptom of accumulated optimisms (CB Insights).

Planning fallacy is not a moral defect: it is a documented cognitive mechanism (we plan for the ideal case and avoid thinking about what can go wrong, because doing so is uncomfortable). The antidote proposed by Flyvbjerg's research is to change the reference point: plan from the distribution of outcomes of similar projects (how comparable cases typically end), not from the optimistic details of your own plan.

The common pattern: the four signatures of a costly mistake

The four cases share signatures that any executive can learn to detect before signing off on the decision:

None of these questions requires being a financial genius; they require someone with authority to ask them before the decision, and a team that does not treat them as heresies. In the four stories above, at some point someone could have asked them. The difference between a six-figure mistake and a timely correction is rarely information: it is institutional permission to raise the uncomfortable question.

How to apply the anatomy in your company: the anti-mistake memo

The practical instrument to close this series is a one-page memo that accompanies every high-impact decision, with five fields:

1. The decision in one sentence and what alternatives were ruled out (forcing real alternatives to exist).
2. The failure scenario: what this poorly made decision would look like, and what early signal would indicate it (the "optimistic planning" signature is fought by naming failure in writing).
3. The cheapest possible test to validate critical hypotheses before full commitment (the Quibi lesson).
4. The point of no return: how much is lost if abandoned in month 3, and whether that exit is clean (the two-way door logic we used in the founder's article).
5. The review date with criteria: what results, in what timeframe, would invalidate the decision (the closing discipline of this series).

No framework eliminates business risk, and this article does not promise to do so. What the anatomy of mistakes promises is something more modest and more useful: that when your decision seems perfectly reasonable, you know the four most documented ways that same reasoning ended badly for others. Kodak, Blockbuster, and Quibi were not poorly managed companies run by negligent people: they were well-managed companies making decisions that (at the time) could be defended. That is exactly the point where history deserves to be studied.

Sources consulted

1. Flyvbjerg, B., Project database (Oxford): statistics on budget, schedule, and benefit compliance. Accessed October 14, 2026: https://budgetoverrun.com/studies/flyvbjerg-megaproject-database
2. Penguin Random House. How Big Things Get Done (Flyvbjerg & Gardner, 2023), official book page. Accessed October 14, 2026: https://sites.prh.com/how-big-things-get-done-book
3. Forbes (2012). How Kodak Failed (Chunka Mui, January 18, 2012). Accessed October 14, 2026: https://www.forbes.com/sites/chunkamui/2012/01/18/how-kodak-failed
4. Rochester Business Journal (2017). Kodak's decades of decline (September 13, 2017). Accessed October 14, 2026: https://rbj.net/2017/09/13/kodaks-decades-of-decline
5. CNBC (2020). How Netflix almost lost the movie rental wars to Blockbuster (September 22, 2020). Accessed October 14, 2026: https://www.cnbc.com/2020/09/22/how-netflix-almost-lost-the-movie-rental-wars-to-blockbuster.html
6. CNBC (2020). Quibi to shut down after just 6 months (October 21, 2020). Accessed October 14, 2026: https://www.cnbc.com/2020/10/21/quibi-to-shut-down-after-just-6-months.html
7. CB Insights. Why Startups Fail: Top 9 Reasons (analysis of 431 post-mortems). Accessed October 14, 2026: https://www.cbinsights.com/research/report/startup-failure-reasons-top
8. Startup Genome. Premature Scaling: A Deep Dive. Accessed October 14, 2026: https://startupgenome.com/insights/premature-scaling-a-deep-dive