Insights

Anatomy of a failing capability center, and how to rescue one

A capability center fails in instalments, and the cause is almost always its original design rather than its people. The rescue works only if it first names the failure mode, then follows one sequence: stabilise delivery, rebuild skills, then transfer decision authority.

This summarises Chapter Seven of The Micro-GCC Advantage by Anupam Shrivastav (2026). The figures are the book's own, drawn from the research it cites.

Three ways a center fails

The book describes three failure modes. They share a root cause, a governance deficit, but they show up differently and need different fixes.

• Delivery failure. Deadlines slip, quality drops and escalations rise. The usual root cause is a scope mismatch from the start: the center was designed for version one of the problem and is being asked to solve version three.
• Skill failure. The center has capacity but not capability. It hired capable people, then gave them work that never needed their full ability, so the ceiling became theirs.
• Governance failure. The work is adequate but nothing of strategic weight happens. The authority to make real decisions was never handed to the center. The book cites research that 92 per cent of underperforming centers keep decision-making highly centralised at headquarters.

The three rarely appear alone. Fixing only one gives a short improvement and then a relapse.

The rescue sequence

The book calls the order load-bearing: each step creates the conditions for the next.

1. Establish an honest baseline. Diagnose all three failure modes with the center's most capable mid-level people involved, not just its leadership or headquarters. The step is done when the root causes are agreed in writing.
2. Stabilise delivery. Do it without cutting scope. Handing work back to headquarters stabilises the metric but is a retreat, and it will later be used as proof that the center cannot cope. The book's completion signal is service levels at or above standard for three consecutive months without scope reduction.
3. Reset the talent architecture. Make permanent senior hires, redesign development around outcomes, and redesign the best people's roles to need and reward depth. The book cites research that 50 per cent of top-performing centers have locally tailored employee value propositions, against 8 per cent of underperformers. Its signals: senior vacancies below 10 per cent and more than half of leadership roles filled by internal promotion.
4. Transfer decision authority. Give the center's leaders budget authority within agreed limits, judge the center on outcomes rather than activity, and include its leaders in the parent company's strategic planning. Without enforcement the book expects this to revert within twelve months.
5. Build irreplaceability. Pick two or three domains where the center can build real depth. The book's test is that the parent company names the center its primary center of excellence in at least two strategic domains.

The book puts the timeline at eighteen to thirty-six months for a center of average size, depending on how long the failure was allowed to compound. Compressing it, it argues, produces nominal rather than genuine completion.

First ask whether it is worth rescuing

Not every failing center is. The book asks three questions answered with evidence, not hope: is the talent still there, can the city support the mandate, and is headquarters genuinely prepared to change? If not, a rescue will consume the budget and credibility better spent elsewhere. The wider decision is covered in build, fix or walk away.

The simplest fix is often authority

The book describes a European pharmaceutical company whose capable regulatory center in India was stuck because approval authority stayed in Europe. Work waited for sign-offs the centre could not give. A documented transfer of approval authority for specified document categories ended the duplicate effort and restored morale, with no new talent or technology. In the book's words, the capability was always there, waiting for the permission.

Related: why the cost case alone fails, and the glossary. This is general information, not management or financial advice.

Questions on this topic, or a situation you are weighing? Write to hello@altacentpartners.com.