Most discussions of enterprise architecture maturity start with a five-level model and an arrow pointing at Level 5. Useful for a slide deck. Less useful for the architect on Monday morning who can't get a clear answer to «which applications depend on the legacy ERP we're decommissioning?»
The honest version: most organizations are at Level 2, a few have made it to Level 3, almost no-one is genuinely at Level 4 - and that's fine. Maturity is a diagnostic, not a destination.
The five levels in one paragraph each
Level 1 - Initial. No formal EA practice. Architecture happens by project, by accident, and by tribal knowledge. Decisions repeat themselves because no-one remembers the last time they were made. «Which app owns this data?» gets you a meeting, not an answer.
Level 2 - Developing. Someone wears the EA hat. There's a Confluence space, maybe a Visio file, maybe a written list of standards. But the catalog of applications, integrations and data is partial, drifts within a quarter, and gets reconstructed for every audit. Most organisations live here.
Level 3 - Defined. A real catalog exists, owned, kept current. Architecture reviews happen on a schedule. Changes go through some kind of approval. The catalog answers questions like «which apps touch confidential customer data?» without a fire drill. Getting from 2 to 3 is the hardest jump on the ladder.
Level 4 - Managed. EA is part of project initiation, not bolted on after the fact. Metrics are tracked: capability coverage, integration counts by lifecycle stage, technology sprawl. Decisions reference the model. The model is the source of truth - not a copy of it.
Level 5 - Optimising. EA shapes business strategy as an input, not as a check-the-box deliverable. Tech investments are deliberate. Continuous improvement is everyone's job. This level is rare and hard to sustain - it depends as much on organisational culture as on architecture practice.
How to tell where you actually are
Forget the levels for a moment. The honest test is one question:
If your CIO asks «which integrations carry PII into our reporting layer, and who owns each?», how long until you have an answer?
Under five minutes - Level 4 or higher. Within the day, by talking to two people - Level 3. End of the week, Slack thread plus a spreadsheet - Level 2. «We'd need to start a discovery project» - Level 1.
For most organisations, the honest answer is Level 2. The catalog technically exists; it just isn't trustworthy or current.
What gets you from 2 to 3
Two things, and only two:
A catalog that's actually current. Not a Confluence page someone updates twice a year. Not a Visio file. A live, queryable model where applications, integrations, data objects and ownership are explicit - and kept current as a side-effect of doing the work. If updating it is harder than not updating it, it will drift, and you'll be back at Level 2 within two quarters.
A change process that goes through it. Architecture decisions reviewed against the model. Changes captured as proposals, not as fait-accompli emails. The model isn't a description of the change - it's the artifact through which the change happens.
Everything else - review boards, ADRs, capability maps, KPIs - is downstream of those two. Skip them and you're decorating Level 2.
Why most organisations stall at Level 2
Three failure modes, all common:
Tool fixation. Buying ServiceNow APM or LeanIX without solving the data problem. The tool isn't the bottleneck; keeping the data current is.
Documentation theatre. Spending six months building a beautiful catalog, then watching it go stale because no-one's job depends on it staying current.
Process without practice. Writing architecture review processes that no project manager has time to follow, then being surprised when no-one follows them.
The pattern in all three: trying to skip the discipline of keeping a model current.
Where Albumi fits
The 2 → 3 jump is exactly what Albumi is built for. Applications, integrations, data objects and capabilities are first-class entities in a single workspace. Changes go through Architecture Change Requests reviewed against the model. The catalog stays current because that's where the work happens - not next to it.
It won't make you Level 5 by itself. It will give you a catalog that survives the second quarter - which is the ladder rung most teams have been failing to climb for years.