Maturity Model
Six stages of architectural capability. Each is defined by what a person decides alone, not by what they can explain.
The distinction matters. Stated knowledge and capability diverge sharply in this profession: it is common for someone to explain eventual consistency well and never have decided, under pressure, whether a concrete case tolerates it.
Two different axes
This model measures capability. The seven levels of the path organize content. They are not the same thing and they do not advance together.
Reading Level 04 puts nobody at stage 4. Content is input; capability comes from deciding, being wrong, recording why, and revisiting. The path shortens the route — it does not replace it.
In practice, someone at stage 3 usually gains more from rereading Level 02 against their own system than from advancing to Level 05.
The stages
Stage 1 — Code-oriented
Decides: how to implement inside an already defined module. Horizon: the current task. Negotiates with: nobody — the scope arrives already cut.
Writes correct, readable code. Applies patterns when the problem is recognized. Does not decide where boundaries sit.
Transition signal: starts noticing that the difficulty of a task comes from an earlier structural decision, and can name it.
Stage 2 — Design-oriented
Decides: module structure, interfaces and direction of dependency. Horizon: the feature, a few weeks. Negotiates with: their own team.
Draws boundaries inside a system. Justifies why a dependency points one way. Recognizes structure degrading before the cost shows up in the roadmap.
Transition signal: starts hitting limits that are not about code — the database, the queue, the deployment process.
Stage 3 — System-oriented
Decides: components, contracts, storage and the topology of a system. Horizon: the system, a few quarters. Negotiates with: product and neighbouring teams.
Goes from requirements to a high-level architecture with capacity estimates. Identifies the bottleneck before the system exists. Chooses between synchronous and asynchronous with an argument.
Transition signal: starts making decisions whose cost lands on other teams, and notices that a technical argument alone is not enough.
Stage 4 — Architecture-oriented
Decides: trade-offs between quality attributes, under cost and schedule constraints. Horizon: the system and its evolution, one to two years. Negotiates with: technical leadership and business stakeholders.
Derives architecture from stated numbers — SLO, RTO, budget — rather than from preference. Records decisions with the context that makes them revisable. Can say no to complexity that does not pay for itself, including complexity they proposed earlier.
Transition signal: begins to see duplication and incoherence between systems, not inside one.
Stage 5 — Enterprise-oriented
Decides: portfolio, capabilities, standards, and the route from current to target state. Horizon: the set of systems, two to five years. Negotiates with: whoever controls budget and priority.
Reads an application landscape and identifies duplicated capability. Builds transition architectures in steps that deliver value on their own. Establishes standards with an exception path.
Transition signal: notices that the dominant constraint has stopped being technical and become organizational.
Stage 6 — Strategy-oriented
Decides: technical direction, team structure, and where the organization invests capacity. Horizon: the organization, years. Negotiates with: company leadership.
Treats Conway's law as an instrument, not an observation: proposes organizational change to make an architecture possible. Measures architectural outcomes instead of arguing for them. Sustains coherence without centralizing decisions.
How to use this
Locate yourself with one question: what was the most consequential decision you made alone in the past six months, and what kind of decision was it?
The answer is usually one stage below what self-assessment suggests, because we tend to measure ourselves by what we understand rather than by what we decide.
Two warnings. There is no merit in being at stage 6 — there is fit to a role; a thirty-person company rarely needs one. And the stages are not exclusive: someone at stage 5 still makes stage 2 decisions, and still needs to make them well.