A decision left open in design is settled informally in operation, usually more than once and in more than one way.
Why it matters when the plan changes
Design phases end when the structure is agreed, which is earlier than when the decisions are agreed. A target operating model is, by definition, a statement of intent describing a desired state, not a record of every decision that state will require. Questions that could not arise under the old model are absent from anyone's mind when the new one is drawn, so they stay unspecified. The first person who meets each question settles it locally, and unwinding that later costs authority as well as time.
The tension is that a design phase extended until every question is answered never ends, and the strategy moves while it runs. Cross-unit commitments show the cost: Sull, Homkes and Sull found only nine per cent of managers could rely on colleagues in other functions all the time, which is what an unfinished design leaves to chance. The workable discipline is coverage of the decisions the plan depends on, a shorter list than the set the model will ever require.
In practice
A target model defines clear accountabilities on one page and is signed off. It does not say who settles a pricing conflict between a product line and a market, because under the old structure that question could not arise. Within a quarter it arises weekly and is settled differently each time, by whoever escalates fastest.
Evidence
A target operating model describes the desired state, produced against the current one, and is a statement of intent.
Target operating model, Wikipedia (2026)Commitments across units are the least reliable part of execution, and they are what an unfinished design leaves to chance.
Donald Sull, Rebecca Homkes and Charles Sull, Why Strategy Execution Unravels and What to Do About It, Harvard Business Review (2015)
What it cannot tell you
The distinction doesn't tell you which decisions matter enough to need settling in design, only that leaving them open costs more later. It also doesn't specify how to detect that a question exists before the new model surfaces it, so the framework can only be applied retrospectively.
Questions
Formally when the structure is signed off; the Target operating model entry on Wikipedia, updated 2026, describes this as a statement of intent, not a record of decisions. A better test is whether the plan's contested calls have named owners. By that test, many design phases end months after they are declared complete.
The ones the plan depends on and the ones the new structure makes possible for the first time. The second category is the one that gets missed, because questions that could not previously arise are not in anyone's mental model when the model is drawn.
They are settled informally by whoever needs an answer first, often differently in different places, the same cross-unit gap Sull, Homkes and Sull measured in Harvard Business Review in 2015, finding only nine per cent of managers could rely on colleagues in other functions all the time. Formalising the answer later costs authority as well as time.
Long enough to cover the decisions the plan depends on and no longer. Extending it until every question has an answer means the strategy moves underneath it, and the model ends up designed for a plan that has already changed.
They usually do, and that is manageable when it is deliberate. What is not manageable is operating under a model whose contested decisions were never assigned, because the organisation then converges on informal answers faster than the design can catch up.