Agency vs. Luck: An Honest Accounting

Both things are true: your decisions shaped the outcome, and forces outside your control shaped it too. The honest founder holds both propositions simultaneously — not as a philosophical exercise, but as a practical discipline that produces better strategy, more accurate post-mortems, and a more durable operating identity. Crediting luck accurately does not make you a weaker operator; it makes you a more calibrated one.

Why founders get this wrong — and why it costs them

The entrepreneurial canon is built on agency. Founders are, by self-selection, people who believe their actions change outcomes. Research confirms that people who believe they have more control over their environment are more likely to undertake a risky endeavor such as entrepreneurship — and that entrepreneurs with a strong internal locus of control are less likely to believe their success is attributable to luck. This is not a flaw. It is the psychological architecture that makes founders willing to start at all.

But the same architecture that makes founders act also makes them misread results. The Fundamental Attribution Error — the cognitive bias of over-attributing organizational outcomes to leadership dispositions — produces a corresponding under-attribution of outcomes to circumstantial factors. In plain terms: when a company wins, the founder credits the strategy; when it loses, the founder blames the market. Both moves are partially correct and systematically incomplete.

The cost is not merely philosophical. A founder who cannot distinguish what she controlled from what fortune handed her will build the wrong playbook for the next venture, give investors a distorted signal about her edge, and — most consequentially — fail to engineer the structural conditions that actually drove the win. She will try to replicate the narrative rather than the mechanism.

Nassim Taleb’s central argument in Fooled by Randomness is simple but profound: many of the things we see as “skill” are actually lucky breaks, and many failures might not be due to bad judgment at all. He illustrates this by looking at traders, investors, and business leaders who rise to the top only to fall later — often because randomness played a bigger role than they realized. Taleb is not arguing that skill is irrelevant. He is arguing that we are systematically unable to see the counterfactual histories — all the equally skilled operators who ran the same play and lost because the macro wind shifted.

Because of hindsight bias and survivorship bias, we tend to forget the many who fail, remember the few who succeed, and then create reasons and patterns for their success even though it was largely random. Mild success can be explainable by skills and hard work, but wild success is usually attributable to variance and luck. This is the distinction that matters most for a founder doing an honest post-mortem: was this a 2x outcome or a 20x outcome? The answer changes the attribution calculus entirely.

What the data actually says about timing, structure, and the founder’s hand

The most rigorous practitioner dataset on this question comes from Bill Gross, founder of Idealab. To test his assumptions, Gross systematically analysed 200 companies — 100 from Idealab and 100 external companies — ranking each on five startup success factors and comparing massive successes like Airbnb, Instagram, Uber, YouTube, and LinkedIn against notable failures including Webvan, Pets.com, and Friendster. The results completely upended his expectations: timing accounted for 42% of the difference between success and failure. Team and execution came second. The idea, which Gross had always considered paramount, ranked third.

This is a striking finding, and it deserves precise interpretation. Timing is not purely luck. Timing refers to the broad condition of the ecosystem — something that is hard to define, and harder to control. You can have an incredible team and a top-tier product, but enter the market too soon and you could be waiting for a penny that will never drop. Enter too late and you may have missed your window. There are ways to mitigate the risk of bad timing through intensive market research and competitor analyses, but it remains one of the most elusive aspects of startup success. Some timing is read correctly through disciplined analysis. Much of it is structural — the macroeconomic moment, the regulatory environment, the infrastructure maturity — and no amount of founder skill moves those variables.

Consider the canonical cases. Airbnb’s success was attributed to the perfect timing during the recession, when people needed extra income and were willing to rent out their homes. Similarly, Uber launched at a time when drivers were looking for additional income, making it easier for the company to attract and grow its driver network. Both founding teams were skilled. Both also caught a structural tailwind — the 2008–2009 financial crisis — that no amount of product genius manufactured. The honest accounting credits both.

On the skill side, the evidence is equally clear. A landmark NBER study by Gompers, Kovner, Lerner, and Scharfstein found that entrepreneurs with a track record of success are more likely to succeed than first-time entrepreneurs and those who have previously failed, and that funding by more experienced venture capital firms enhances the chance of success — but only for entrepreneurs without a successful track record. The implication is direct: skill compounds, and experienced VCs can identify it — but only when the founder has not yet demonstrated it themselves. Prior success is a signal that the market reads as skill, regardless of how much luck was embedded in that prior success. Successful entrepreneurs exhibit persistence in selecting the right industry and time to start new ventures, and those with demonstrated market timing skill are more likely to outperform industry peers in subsequent ventures. The skill of reading timing, in other words, is itself learnable — which means it migrates from the luck column toward the agency column over a career.

The survivorship problem: why the sample lies to you

Any honest accounting of luck must grapple with the sample from which founders draw their lessons. The narratives that dominate the discourse often fail to shed light on a hidden menace that skews our perceptions of success — survivorship bias. Survivorship bias occurs when we concentrate only on the entities or individuals that have succeeded while ignoring those that have failed or dropped out along the way. This cognitive fallacy can lead to inaccurate assessments, misinterpretations, and misguided decisions.

The problem is structural. A study analyzing transcripts derived from 183 podcast interviews that ask successful entrepreneurs whether luck or skill accounts for their success found that founders’ attributions differed by gender, while race and geography of birth showed no strong difference. The sample, however, is already filtered: only founders who survived long enough to be invited onto podcasts are in the dataset. The founders who ran the same playbook and failed are not being interviewed. Their absence from the discourse is not evidence that the playbook was wrong — it is evidence that the sample is biased.

It is natural for those who failed to vanish completely. Accordingly, one sees the survivors, and only the survivors, which imparts a mistaken perception of the odds. A founder who studies only successful companies is studying a population that has already been filtered by a combination of skill, timing, capital access, network, and fortune. Extracting clean lessons from that population requires actively reconstructing the counterfactual — the equally capable founders who did not make it.

This is not an argument for paralysis. It is an argument for epistemic precision. The founder who knows she is drawing lessons from a survivorship-biased sample will weight those lessons differently. She will ask: “Would this principle have worked in a down market? Would it have worked without the specific network advantage I had? Would it have worked if we had launched eighteen months earlier?” These are not rhetorical questions. They are the questions that separate a replicable system from a lucky narrative.

The two errors: overclaiming agency and abandoning it

There are two failure modes in this accounting, and they are not symmetric in their consequences.

Error one: overclaiming agency. This is the dominant error in founder culture. One of Taleb’s most striking observations is how we naturally overestimate our ability to control outcomes. When things go well, we attribute it to skill. When things go badly, we look for external excuses. The founder who overclaims agency builds a mythology around her decisions that is not falsifiable. She cannot learn from the luck that helped her, because she has reclassified it as skill. She cannot prepare for the luck that may not recur, because she believes she can manufacture it. And she will almost certainly misattribute her next failure — blaming execution when the real variable was a macro shift she had no hand in.

Error two: abandoning agency. The opposite error is less common among founders but more common among observers who have absorbed the luck literature too literally. If timing accounts for 42% of variance, the conclusion is not that the other 58% is irrelevant — it is that the other 58% is where the founder actually operates. Research suggests that people who believe they have more control over their environment are more likely to undertake a risky endeavor such as entrepreneurship. The internal locus of control is not a delusion to be corrected; it is a functional prior that enables action under uncertainty. The goal is not to replace it with fatalism. The goal is to calibrate it.

Research on locus of control in small business settings suggests that internal locus of control indirectly affects venturing outcomes via entrepreneurial competency, whereas external locus of control has no such consequences. In other words, believing you have agency is instrumentally necessary for developing the competencies that produce results. The founder who has fully internalized that luck dominates will not build the systems, develop the skills, or make the bets that compound over time. She will wait for fortune to arrive rather than positioning herself to capture it when it does.

The synthesis is not a compromise between these positions. It is a more precise model: agency determines your surface area for luck; luck determines which of your positioned bets pays off. The founder’s job is to maximize the former while remaining honest about the latter.

What accurate luck-accounting actually looks like in practice

An honest post-mortem on any venture outcome should produce a ledger with two columns. The agency column contains decisions that were made with information available at the time, that were causally connected to the outcome, and that a different operator might plausibly have made differently. The structure column contains macro conditions, timing variables, network effects inherited rather than built, and capital market dynamics that were not within the founder’s control.

Several practical disciplines support this accounting:

• Reconstruct the counterfactual. For every outcome you are tempted to credit to a decision, ask whether a founder with identical skills but a different macro environment would have achieved the same result. If the answer is probably not, the macro environment deserves a share of the credit.
• Separate timing from execution. Work hard to understand the PESTEL factors that positively or negatively impact your markets. Understand what is inside your control or sphere of influence, and understand what you purely need to monitor. This is not a strategic planning exercise — it is an attribution exercise. Knowing which variables you monitored versus which you moved is the foundation of an honest ledger.
• Weight the magnitude of the outcome. Mild success can be explainable by skills and hard work, but wild success is usually attributable to variance and luck. A 3x return on a venture is a different attribution problem than a 300x return. The larger the deviation from the base rate, the more seriously you should interrogate the luck hypothesis.
• Audit your network inheritance. Access to capital, warm introductions to early customers, and the credibility that comes from a prestigious prior employer are structural advantages that compound silently. They are not evidence of bad faith — but they are not evidence of skill either. Naming them accurately is the beginning of building systems that do not depend on them.
• Apply the same standard to failure. We typically hold an internal locus of control for our successes and an external locus of control for our failures. The honest accounting runs in both directions. The venture that failed because of a macro shock deserves the same structural analysis as the one that succeeded because of a macro tailwind. Asymmetric attribution — skill for wins, bad luck for losses — is the most common form of self-deception in the founder population.

None of this requires public confession. The honest accounting is primarily an internal discipline — a private ledger that informs how you build the next system, how you advise the next founder, and how you evaluate the next opportunity. It does not require you to qualify every success story with a disclaimer. It requires you to know, privately and precisely, what you actually did.

The compounding benefit is strategic clarity. A founder who knows that her first exit was 60% timing and 40% execution will not try to replicate the timing — she will try to replicate the execution and position herself to catch the next timing window. A founder who believes the exit was 100% execution will try to run the same play in a different market cycle and be confused when it does not work. The honest accounting is not modesty. It is calibration. And calibration compounds.

This connects directly to how credibility is built over time — a subject explored in depth at /history-of-credibility/ and /how-trust-develops/. Investors and partners who have seen many founders operate can often distinguish the ones who have done this accounting from the ones who have not. The tell is not humility — it is precision. The founder who can say “we won because of X, and we also caught Y tailwind that we cannot guarantee will recur” is demonstrating a diagnostic capability that is itself a signal of operator quality. It is also the foundation of the kind of /how-a-deal-closes/ credibility that closes rounds at favorable terms.

What this means

The forward position

The founder who has done this accounting honestly arrives at a more durable operating identity than the one who has not. She knows what she is actually good at, because she has separated it from what she was lucky enough to catch. She knows which structural conditions she needs to recreate, because she has named them rather than absorbed them silently into her mythology. And she knows which variables she cannot control, which means she can build systems to monitor them rather than pretending she can move them.

This is not a counsel of modesty. It is a counsel of precision. The most capable operators in any market are not the ones with the most confident narratives about their own genius — they are the ones who can read a situation accurately, including the parts of it that have nothing to do with them. That capability is what compounds. The narrative does not.

Agency is real. Structure is real. The founder who holds both propositions simultaneously — and can tell the difference in a specific situation — is operating at a level of clarity that most of her peers will never reach. That clarity is itself a competitive advantage. It is also, for what it is worth, the truth.

For those building the systems and verification infrastructure that make structural conditions legible — including how capital flows to founders across different market contexts — the analysis at /capital-platforms-developing-economies/ and /alternative-credit-data/ extends this framework into the structural conditions that shape which founders even get a surface area for luck in the first place.