Ries lists three obvious answers to the question “what is a startup for?”—making a product, making money, and serving customers—and rejects all three in favor of a fourth: learning how to build a sustainable business. This is a teleological move: learning here is neither a means nor a side effect, but the purpose of existence. Learning is validated scientifically through frequent experiments that test elements of the vision (validated learning); “scientific” here means falsifiability and objective behavioral metrics, not laboratory rigor. [ET] Validated knowledge is a means of observable uncertainty reduction: observability distinguishes validated learning from self-persuasion (“we learned a great deal”) and makes learning demonstrable—progress that can be shown to those who hold the innovator accountable (a bridge to innovation accounting, principle 5).
The formulation in Chapter 1 states the same telos, compressed to the product and supplied with pace: a startup’s goal is to determine the right product (the right thing to build—what customers want and will pay for) as early as possible. Two differences from the introductory formula are worth retaining. First, the narrowing: “the right product” is narrower than “a sustainable business” (model, customers, monetization); the product is a special case, and the Chapter 1 formula must not replace the broader one. Second, speed becomes part of the goal: not merely to learn, but to learn quickly—hence the feedback loop as the method’s central mechanism (cf. Fast feedback - a universal advantage).
Why this move is load-bearing—three direct consequences:
- The unit of progress changes: startup progress is validated knowledge, not features, releases, or revenue. From this follows directly the need for new accounting (innovation accounting, principle 5): traditional reporting does not register this kind of progress, while “vanity metrics” merely simulate it.
- The product’s status is reversed: if the goal is learning, then the product is a vehicle for an experiment, a question posed to the market rather than an answer. An early poor MVP ceases to be a disgrace and becomes a tool of inquiry.
- The cost of failure is recalculated: an unsuccessful experiment is knowledge gained, not a loss; only work that produces no knowledge becomes a loss (waste in the lean sense).
Boundaries: the link to commercial success is instrumental, not severed—learning is undertaken about business—but research success is not equivalent to company survival: one can learn exemplary lessons about the fact that a business is dead. The teleology makes the strong claim that “success can be constructed” defensible at the cost of weakening it almost to a tautology—the gap between the promised commercial success and the guaranteed research success remains part of the book’s rhetoric.
How it supports Startup success is constructed - the right process can be learned
It supplies the reading of “success” under which the claim of constructibility is defensible: a process reliably produces validated knowledge, but not a commercial result. Without this redefinition, the claim of constructibility is vulnerable to the counterexample “perfect process—dead product.”
Parallels in the repository (connections made here):
- The result of a project is not a working product, but the situation in which it is used—the same movement of thought on the personal scale: the outcome of activity is redefined through a higher-level telos, while the artifact is reduced to a carrier of value. For Ries, the product is a carrier of an experiment; there, the working product is a carrier of future situations of use. The pragmatics match too: “the existence of situations of use as a filter for projects” is analogous to a falsifying experiment as a filter for features.
- The Fifth Discipline, The Five Disciplines of Learning Organizations, Formulating a Definition of a Learning Organization—Ries’s startup is Senge’s learning organization taken to the limit: learning is not one discipline among others, but the telos itself. The definition exercise already contains features of a startup in Ries’s sense: “people are free to experiment, take risks, and openly assess results,” “the jungle of ideas… experiments first—we use the knowledge that is harder to criticize.”
- Adaptation to the environment depends on the objectivity of perception and Objectivity as the possibility of falsification—ground the word “validated”: objectivity = admissible falsifiability (Schön). A non-obvious implication follows: bad mental models survive where the cost of error is too low for them to be culled. For a startup, the cost of error is instead maximal (the company’s death), but delayed—and therefore it does not teach. Validated learning builds in an artificial selection mechanism: cheap, quick experimental refutations instead of one expensive late refutation by the market.
- Fast feedback - a universal advantage—adjacent: the telos determines what to maximize (validated knowledge); the Build-Measure-Learn loop and OODA determine how (the speed and frequency of feedback as a competitive advantage).
- Strategizing (ShSM)—the same telos on the personal scale: strategy as a constant cycle of proposing and testing hypotheses; a personal project exists in part to test a strategic hypothesis.