From Best Practices to First Principles in Organizational Learning

When organizations decide to get better at learning, they usually reach for best practices. They benchmark admired firms, import the frameworks that worked elsewhere, and adopt the routines that a case study credited with success. The instinct is understandable and occasionally useful. But it rests on a mistake that these essays have been circling from the start: the belief that a practice which succeeded in one place carries its success with it, when what it actually carried was an answer fitted to a particular question that may not be yours.

The research on this is unusually clear-eyed. Szulanski's study of the internal transfer of best practices found the process to be surprisingly sticky—that even within a single company, moving a proven practice from one unit to another is hard, and that the stickiness comes largely from missing context: the tacit knowledge, the local conditions, the surrounding routines that made the practice work in its original home. The contingency tradition made the same point at the level of theory decades earlier, in the deceptively simple finding that there is no one best way to organize; what works depends on the conditions. A practice is an answer entangled with its circumstances. Strip away the circumstances and you are left with a form, not a cause.

This is why best-practice adoption so often disappoints. The organization copies the visible routine—the ritual, the template, the ceremony—without the invisible conditions that gave it force, and is puzzled when the imported form produces none of the promised effect. It has learned the wrong lesson in the specific sense described earlier in this series: it has attributed a success to its most copyable feature rather than to its actual cause. Benchmarking, done this way, is superstitious learning conducted at industrial scale, one organization borrowing another's coincidences.

There is a subtler cost as well. A collection of imported best practices is not a coherent whole. Each was extracted from a different context and optimized for a different end, and assembled together they need not fit—the measurement practice borrowed from one firm may pull against the learning culture borrowed from another, and the knowledge-management system copied from a third may quietly undermine both. This is the fragmentation these essays have described, arriving now by a particular route: an organization can accumulate a drawer full of individually admired practices and possess no working system, because it assembled parts without a principle to make them cohere.

The alternative is to reason from first principles. Rather than asking what successful organizations do, ask what the problem actually requires: what must be true of any organization that turns experience into reliable, current, tested capability? What does learning need in order to be valid rather than superstitious? What does memory need in order to help rather than ossify? How must measurement, decision, and adjustment relate for the whole to correct itself rather than drift? These questions do not have their answers lying around in someone else's case study. They have to be worked out from the nature of the problem—and their answers, unlike borrowed practices, can be made to cohere, because they are derived from a single account rather than scavenged from many.

First-principles reasoning has a further advantage that borrowing can never offer: it travels. A practice copied from another firm is bound to the conditions it came from and breaks when those conditions differ, which is why benchmarking works so erratically. A principle derived from the nature of the problem applies wherever the problem takes that nature, and adapts as conditions change, because it specifies not a fixed routine but the requirement a routine must satisfy. This is the difference between being handed a fish and understanding why fish live where they do. An organization armed with principles can generate the practice its own situation calls for; an organization armed only with practices can copy, and hope its situation resembles someone else’s closely enough to survive the transfer.

Reasoning from first principles is harder and slower than copying, and it forfeits the comfort of pointing to someone who has already succeeded. But it is the only route to a coherent whole, because coherence cannot be imported piecemeal; it has to be built in from the start, as a property of the design rather than a happy accident of assembly. The organizations that will handle the conditions described in this series are unlikely to be the ones with the best-stocked drawer of borrowed practices. They will be the ones willing to ask what those practices were dim reflections of—and to build, from the principles up, the coherent system that no amount of benchmarking can hand them. What such a system must satisfy to be worth building is the question the final essay takes up.