Effectuation, Twenty-Five Years Later
In 2001, the web startup I had co-founded two years earlier was dying at roughly the speed of the NASDAQ. We had a business plan. A proper one: market sizing, growth curves, revenue projected three years out with two decimals of confidence. Somewhere between the second and third tab of that spreadsheet, the market it predicted stopped existing.
I did what autodidacts do when the ground moves. I went looking for better sources.
That same year, a researcher named Saras D. Sarasvathy (a student of Herbert Simon, which should have been the first clue that this was not ordinary management literature) published a paper with a title only an academy could love: Causation and Effectuation: Toward a Theoretical Shift from Economic Inevitability to Entrepreneurial Contingency. I read it in the wreckage of my own growth curves. It did not read like theory. It read like an autopsy report on my company, written before the death.
The paper, in one fridge
The distinction Sarasvathy drew is simple to state and slow to digest.
Causation is the logic every business school taught then, and most still teach: take a goal as given, then assemble the optimal means to reach it. Pick the market, size it, plan backwards from the prize. The spreadsheet.
Effectuation inverts the arrow. Take your means as given (who you are, what you know, whom you know) and let the goals emerge from what those means make possible. My own shorthand, then and now: stop shopping for a recipe. Cook from what is in the fridge.
Her running example was an imaginary Indian restaurant called Curry in a Hurry. The causal entrepreneur studies the restaurant market of an entire city and raises money to build for it. The effectual one starts cooking for the people she can already reach, watches what they actually want, and lets the business become whatever the responses make of it. Maybe a restaurant. Maybe a catering line. Maybe something no market study would have listed.
The paper then names four principles, and I want to quote their logic faithfully because the cute labels came later: affordable loss rather than expected returns; strategic alliances rather than competitive analyses; exploitation of contingencies rather than preexisting knowledge; and control of an unpredictable future rather than prediction of an uncertain one. Years afterward, the pedagogy dressed them up as bird in hand, crazy quilt, lemonade, pilot in the plane. The 2001 paper is drier and better.
I was twenty-five years younger, freshly back from four years of deploying power plants on the Indian subcontinent, and I had just watched prediction fail at scale. I thought I understood the paper.
I understood maybe a third of it. The rest took four companies.
Starting from what was in the fridge
The startup I lost in 2001 was pure causation. We had chosen a market the way you choose a destination on a map, then went looking for the means to get there. When the map burned, we had nothing left, because nothing we had built started from us.
OM Conseil, founded in 2003, started the other way. The inventory of means was short and honest. Who I was: a technician who had learned management the hard way, in another hemisphere. What I knew: how to keep infrastructure running for organizations too small to staff the skill themselves. Whom I knew: a handful of people who trusted me enough to pay for that. That was the fridge. The firm that grew to thirty-five people over nineteen years never came from a market study. It came from cooking what those first ingredients allowed and paying attention to who came back for more.
I have told the story of what happened inside that company in Nineteen Years Inside a Sociocratic SMB. What I want to add here is the part effectuation explains: the firm’s best turns were never in any plan I wrote.
The loss I could afford
The 1999 startup was sized on expected returns. We committed what the projections said the upside justified, which is another way of saying we bet money we did not have on a future we did not control.
In 2003 the question changed. Not “how big can this get” but “what can I stand to lose if it fails”. The founding of OM Conseil was sized to that answer: small office, no debt, my own time as the main stake. It made the first years slower than a funded competitor’s. It also meant that no single bad year could kill us by itself.
Twenty years on, this is the principle I use most, and not only for founding things. Every small tool I ship as an indie hacker is an affordable-loss bet: a few days of my time, infrastructure that costs less than a restaurant meal, no commitment that survives my decision to stop. When the downside is capped, you can afford to be wrong often. Being wrong often, cheaply, is the closest thing to a method for being right that I know.
The quilt I refused to sew
Sarasvathy’s third principle, strategic alliances over competitive analysis, is the one I ignored longest, and it sent me the most expensive invoice.
In 2009 we lost our largest client, more than half of our revenue. I have written elsewhere about what that collapse revealed. Read through the effectuation lens, the diagnosis is even sharper: I had treated the relationship as a possession to defend instead of a partnership to widen. The client was never woven into the quilt. Neither, for that matter, was my own team, who knew things about that account I never asked. A partner self-selects into your venture and holds a corner of it. A possession just sits there until someone takes it.
The governance transformation we undertook afterward (sociocratic circles, roles instead of titles, decisions distributed) was many things. One of them was this principle, finally applied: turning employees, clients, and even suppliers from resources under management into stakeholders holding corners of the quilt.
Lemonade, twice
Exploitation of contingencies is the principle that sounds the most like a poster and costs the most to practice, because it only ever presents itself dressed as bad news.
The 2009 crisis was the first lemon. Two years of recovery became the occasion to rebuild the company’s entire operating system, and the second decade that followed was unrecognizable from the first. None of that transformation was a goal in 2008. The contingency created it.
The sale of OM Conseil in 2022 was the second. I have called that sale the one true failure of the adventure, and I maintain the word. But the aftermath behaved exactly as the paper predicts: stripped back to bare means (what I know, whom I know, and this time twenty-five years of both), a new venture assembled itself around them. L’Esprit du Net was not a plan B. There was no plan at all, which by then I had stopped mistaking for a problem. Even the advisory practice that now occupies a growing share of my weeks grew from contingency: clients kept asking about generative AI, and I followed the questions rather than a positioning document.
The pilot, not the forecast
The fourth principle is the quietest and the deepest: to the extent that you can control the future, you do not need to predict it.
A large part of my work as a fractional CIO consists of being asked for predictions. Three-year roadmaps. Technology bets. Which vendor will still exist in 2030. I give my best reading, and I hold it lightly, because the honest answer is that nobody knows. What I actually sell is not forecasts. It is control: architectures that stay reversible, contracts with exit clauses, commitments sized so that being wrong is survivable. A company that can change course cheaply does not need to see far.
Readers of Adjacent Possibles will recognize the geometry. Your means define which doors exist; walking to one changes the set. Effectuation is the same insight wearing entrepreneurial clothes: you do not stand in the center of the room predicting doors. You walk, and the walking makes them.
Twenty-five years later, the spreadsheet people still outnumber the fridge people, in startups and in IT departments alike. I no longer argue with them. I just note that of the two companies I built, the one designed backwards from a predicted future lasted two years, and the one grown forwards from available means lasted nineteen.
The best way to predict the future is to invent it. — Alan Kay, Xerox PARC, 1971