intermediate 1 min answer

An enterprise architecture function is asked to cover business, application, data and technology domains. Where should its effort concentrate in a fast-growing company?

ofbusinessea-domainsprioritisationdataintegration
Show the full answer Hide the answer

Where the leverage is in a fast-growing company

Data and integration, disproportionately. Application and technology choices are reversible on a timescale of months; data model decisions, identifier schemes, tenancy keys and integration contracts are not.

Specifically:

  • The core data model and the meaning of shared entities. When three teams disagree about what "customer" or "order" means, the divergence appears in reporting months later and is expensive to reconcile.
  • Identifier and partition schemes, since retrofitting a partition key into a live schema is one of the genuinely painful migrations.
  • Integration contracts between domains, which determine whether teams can move independently.
  • The system of record for each fact, because two authorities for one fact produces permanent reconciliation work.

Where to spend less effort

Technology architecture, beyond the small standardised set with genuine network effects. Most technology choices are reversible and the team bears the consequence.

Application architecture within a team's boundary, which is the team's own decision and where central opinion adds little and costs credibility.

The business domain's role

Not to produce a comprehensive capability map, which becomes a taxonomy exercise, but to answer specific questions: where is investment going, where is duplication, which capability is under-served relative to its importance.

A map built in response to a decision is used; one built as an artefact is not.

The practical prioritisation

In a fast-growing company the EA function is small and the estate is changing weekly. Concentrating on the decisions that are expensive to reverse — data, identity, integration contracts, the regulated perimeter — and explicitly declining the rest is what makes a small function effective.

A function that attempts comprehensive coverage of four domains produces documentation in all four and influence in none.