From Cost Arbitrage to Value Creation: A Framework

Every capability centre that creates real value passed through the same four stages, and most of the centres that never got there stalled at the same boundary. The progression is not automatic, it is not a function of time, and centres regress when the conditions that supported them are withdrawn.

The value of a stage model is that it tells you which problem you actually have. A centre stuck between stage one and stage two needs something quite different from one stuck between stage three and stage four, and applying the wrong intervention wastes a year and some credibility.

This framework sets out the four stages, the exit condition for each, the evidence that you have genuinely passed it, and the typical duration. It also covers what causes regression, because that is more common than most organisations expect.

Stage 1: Arbitrage

The centre executes well specified work. Decisions are made onshore and communicated, work arrives as defined units, and success means adherence to specification at lower cost. Almost every centre starts here and there is nothing wrong with that.

The exit condition is that the centre can complete work correctly without clarification. This is a lower bar than it sounds and most centres reach it within two to three quarters. The trap is that a centre performing well at stage one has no natural pressure to leave, because it is meeting every target it has been given.

The evidence you have passed it: specified work is completed without repeated clarification, and quality is at or above the previous baseline.

Stage 2: Delivery ownership

The centre owns systems end to end and decides how they change. Work arrives as problems rather than specifications, and the centre chooses the approach. This is the largest and most difficult transition in the whole progression.

It is difficult because it requires two things to change simultaneously and neither is comfortable. The seniority mix has to rise, which makes cost per head worse before any benefit is visible. And onshore approval gates have to be removed, which requires named individuals to give up control they are usually being held accountable for.

The exit condition is that the centre resolves ambiguity locally rather than escalating it. The evidence is a falling escalation rate over at least two quarters and a non zero count of systems owned without an approval gate.

This is where most centres stall, and they stall for a specific reason: the transition makes the visible metrics worse for two quarters before it makes anything better. Organisations that evaluate quarterly reverse the change just before it works.

Stage 3: Outcome ownership

The centre owns results rather than systems, and decides what to build in order to achieve them. It is given an outcome, such as a reliability target, a cost reduction or a customer facing improvement, and it determines the work.

This requires the centre to understand the business problem, not only the technical one, which means deliberate investment in context that most organisations never make. It also requires the organisation to tolerate the centre making priority calls it might not have made.

The exit condition is that the centre sets its own priorities within an outcome and the results are accepted. The evidence is that a roadmap area is owned locally and that the centre has declined work on the grounds that it would not serve the outcome, which is a good sign rather than a problem.

Stage 4: Value creation

The centre originates work the business adopts. It identifies opportunities, proposes them, and sees them taken up. At this point it is contributing to strategy rather than executing it.

The exit condition is not really an exit; stage four is a state rather than a stage. The evidence is a non zero and rising count of proposals originated in the centre that the business adopted, and the presence of centre leadership in decisions that used to be made entirely at headquarters.

Very few centres reach this stage, and those that do generally took three to five years. It cannot be reached by skipping stage two or three, because the credibility required for a proposal to be taken seriously is built by having owned and delivered.

The stages summarised

The table below sets out the exit condition and evidence for each stage, which is the practical part of the framework. Durations are typical rather than prescriptive.

The stage two duration is where the variance is greatest. Centres that treat it as a governance change complete it in three quarters; centres that treat it as something that will happen naturally do not complete it at all.

Why skipping stages does not work

Organisations occasionally attempt to establish a centre directly at stage three or four, usually by describing it as an innovation mandate from the outset. This rarely succeeds, and the reason is about the receiving organisation rather than the centre.

A proposal from a team that has never delivered anything is evaluated on its merits alone, which is a much higher bar than it sounds. A proposal from a team that has owned and run a critical system for two years is evaluated with the benefit of established trust. The same idea gets a materially different reception depending on which team it came from.

The one legitimate shortcut is starting at stage two rather than stage one. If the centre is staffed with genuine seniority and given ownership from the first day, there is no reason to spend three quarters executing specifications first. This is the design argument for a small senior centre: it removes the hardest transition by never entering the stage that requires it.

Regression, and what causes it

Progress is not permanent. Centres regress, usually without anyone deciding that they should, and the causes are consistent.

• A new onshore leader reinstates approval gates, often without realising what is being undone
• A serious incident triggers a control response that is never subsequently relaxed
• Senior attrition removes the people whose judgement the ownership rested on
• Cost pressure reduces the seniority mix, which removes the capacity to decide
• Scope is narrowed to fit a delivery emergency and never widened again
• The centre is asked to absorb overflow capacity work, which crowds out the owned outcome

The common feature is that regression is a side effect rather than a decision. Nobody proposes moving a centre from stage three to stage one; it happens through a series of individually reasonable responses to pressure. This is why the stage should be assessed explicitly each year rather than assumed to be stable.

Using the framework

The framework is diagnostic rather than aspirational. Its value is in locating the centre honestly and then applying the intervention that matches.

Frequently asked questions