Innovation.

The framework

1. 01

   Which phase is your category actually in?

   The test is variety. In an open category, competitors differ in kind — different architectures, different assumptions about who buys, different definitions of the job. Once a dominant design lands they differ only in degree: same shape, better execution. Run that comparison across your four nearest rivals and then across your own roadmap; you'll have an answer in a few minutes. If your development spend has quietly shifted from what the product is to how it gets built, delivered and supported, the category's already answered the question for you, whatever your positioning deck still says. That tell looks identical to yesterday's flat top on the S-curve, and often it's the same event seen from two sides: the difference is whether the ceiling sits in the technology or in the category's agreement about what the product is — and only the second one has a date you can point at.

2. 02

   The winning design is chosen socially, not proven technically.

   Anderson and Tushman found the standard settling through licensing terms, supplier commitments, regulation, distribution and the sheer weight of an installed base, rather than through a technical shoot-out. That has a blunt consequence while the category is still open: the work deciding whether your architecture becomes the reference is mostly commercial. It is who else can build on it, who has committed volume, which regulator has written it into a rule, and how many customers would have to re-train if they left. Firms that spend the fluid years perfecting the artefact and none of them recruiting adopters don't lose an argument — they lose a standard.

3. 03

   The morning the skills you hired for stop being the skills you need.

   After a dominant design settles, the returns move to process: yield, reliability, cost per unit, delivery time, the unglamorous work of doing the same thing slightly better every quarter. That is a different organisation. The people who thrived on open questions get restless, the ones who were tolerated as pedants become the most valuable staff you've got, and a founder who keeps reopening the product argument now costs the company money. Most incumbents get caught here twice — once by refusing the shift, and once by making it so completely that they can't do anything else when the next discontinuity arrives.

4. 04

   Keep one bet outside the settled design.

   Anderson and Tushman describe a cycle rather than a destination: a discontinuity opens the argument, a dominant design closes it, incremental improvement runs for years, and another discontinuity opens it again. Knowing that changes when you buy the option, not just whether. The settled era is the cheap window — the operating business is profitable enough to tolerate a small funded position on the architecture that would make yours obsolete, and nobody else is bidding for the people who could run it. Once the ferment starts, everyone can see the same thing and the good ones are already committed. Buy it while it's boring.