A fast-growing technology company is deciding whether to adopt a formal EA framework. What should decide it, and what is the alternative?

What should decide it

The problems the organisation actually has. Formal frameworks solve specific problems well:

• Many business units with overlapping capabilities and no shared vocabulary.
• Regulatory obligations requiring demonstrable traceability from requirement to control.
• A large legacy estate whose interdependencies nobody can enumerate.
• Major transformation programmes where the cost of a missed dependency is enormous.

If the organisation has these, the rigour pays. If it does not, the framework's deliverables age faster than they can be produced, and the function becomes a documentation exercise.

The signal that it is not the right fit: the organisation's actual problems are speed, consistency and coordination between fast-moving teams — for which the frameworks were not designed.

The lightweight alternative

For a fast-growing product organisation, the practices that deliver most of the value:

• Decision records with rejected options and revisit conditions, kept with the code. This alone provides most of the traceability that framework deliverables aim at, at a fraction of the cost.
• A shallow capability map with overlays, deepened only where a decision is pending — enough to spot duplication and to argue investments in business terms.
• A paved road and reference architectures instantiated as templates, which shape the estate far more effectively than standards documents.
• Automated guardrails for the small set of severe, irreversible actions.
• A narrow review trigger set for genuinely significant decisions, with bounded turnaround.
• A small, current technology radar derived from the organisation's own experience.

The multi-market complication

For a company operating across jurisdictions, one framework element is worth adopting early regardless of size: a data architecture that records ownership, classification and residency per entity, propagated through derived datasets.

Retrofitting that is close to impossible, and every privacy, residency and regulatory answer depends on it. It is the one place where the formal discipline earns its cost before scale demands it.

The framing

Adopt the problems, not the framework. List what the organisation cannot currently answer — which systems hold personal data, which capability is duplicated, which decision was made and why — and adopt the minimum practice that answers each. That produces a practice sized to the organisation, which is what tailoring was supposed to mean.