An interviewer says — you run architecture governance at a travel marketplace with four consumer brands acquired over a decade, each with its own stack and its own engineering leadership. The executive team wants one design authority. Where do you take this?

What the interviewer is testing

Whether you can say what a design authority is for without reaching for a meeting cadence. The weak answer designs a board. The strong answer first establishes which decisions must be the same across brands, because that list, and nothing else, determines whether a central body is useful or a tax.

The clarifying questions that change the answer

• What is the unit of consolidation? Front end, booking engine, identity, payments, supplier connectivity and data platform have completely different answers. A brand can keep its own merchandising logic while sharing a payment tokenisation layer.
• Why does each brand exist? If two brands differ only in logo, the architecture question is a portfolio question and the authority is being asked to do someone else's job. If one is a hotel marketplace and another a vacation-rental marketplace with different supply, different fraud profile and different cancellation economics, their domain models genuinely diverge.
• Is there a forcing date? A cross-brand loyalty programme or a regulatory deadline converts an architectural preference into a sequencing problem with a deadline, and that changes everything about how much you centralise first.
• Who holds the budget? An authority with no claim on funding can only block, and a body that can only block gets routed around.

A strong answer's arc

Produce a two-column list before producing a board. Group-level because the decision creates cross-brand coupling or regulatory exposure: identity and account linking, payment tokenisation and card data scope, personal-data residency, the supplier connectivity layer, the front-end platform, the observability and evidence pipeline. Brand-level because the brand's reason to exist lives there: merchandising and ranking, pricing presentation, content model, customer-service policy.

Expedia Group is the documented version of this split. It migrated Hotels.com's front-end stack onto the Brand Expedia platform in 2022 and Vrbo's in 2023, ending with one unified front-end stack across its retail brands, while keeping each brand's distinct role in the market rather than merging them into one marketplace. The consolidation was of the layer where duplication cost the most and differentiation mattered least.

Then, and only then, design the function: a published in-scope list, a decision service level in working days, written decisions with review dates, and an explicit statement of what it does not review.

Common weak answers

• "One platform for everything." It maximises the coupling that acquisitions were supposed to avoid, and the migration cost lands before any benefit does.
• "Start with a canonical data model." The most expensive and least reversible thing you could attempt first, and the one most likely to be abandoned half-finished.
• "A standards document." A ruling nobody can implement is not a decision. Each group-level ruling needs a paved path shipped with it.

What a strong answer adds

A reversal cost next to every item on the group-level list, because that is what justifies central control: centralise what is expensive to un-decide, and leave everything else alone unless it creates cross-brand coupling. A measurement plan, with per-brand delivery lead time recorded for 90 days before the change so the authority can be judged on something. And the honest admission that a central authority raises coordination cost on the brands that were already fast, which is the bill the executive team is signing for. The way this fails in practice is not rejection of good designs; it is teams scoping work to stay below the authority's threshold, which leaves the governance function reviewing the decisions that matter least.