Which enterprise architecture domain produces the most value in practice, and why?

What is being tested

Whether you can identify where enterprise-level work is genuinely irreplaceable rather than merely conventional.

The reasoning

Each domain has value, but three of the four have owners closer to the work:

• Application architecture is largely owned by the teams building the applications; the enterprise contribution is the portfolio view, which is real but narrow.
• Technology architecture is largely owned by platform and infrastructure teams.
• Business architecture is owned by the business, and enterprise architects contribute interpretation rather than authority.

Data is the exception. Data ownership, definitions, lineage and quality problems cross every team boundary by construction, and no single team can resolve them because each sees only its own portion:

• Two teams computing "active user" differently, both correct within their own scope.
• A customer entity written by six systems with no agreed master.
• A number in a board report that nobody can trace to its source.
• Personal data whose location nobody can enumerate, which makes deletion and residency obligations unanswerable.

Each of those requires someone with an enterprise view and the standing to arbitrate. That is the enterprise architect's genuinely distinctive contribution.

The general failure across all four domains

Producing models nobody uses. The four-domain frame is a way of ensuring nothing is overlooked, not a mandate to document everything in each.

The corrective question for any artefact: which decision does this inform, and who is making it? If there is no answer, do not produce it. An EA practice that has produced comprehensive models of all four domains has usually spent its budget on documentation rather than on decisions.

Where the value actually concentrates

At the intersections: a capability implemented by four applications; a data entity owned by nobody and written by six systems; a technology standard that makes a required capability uneconomic.

Those findings are only visible if the domains are related to each other, which is the actual reason to have the frame at all.

What a strong answer adds

That data architecture's value is in ownership and meaning, not in schemas. A catalogue of tables is inventory; a statement of what each field means, who owns it, and how it is derived is what resolves the disputes.