TRANSFORMATION AT SCALE AND SPEED - PART 2

Part 2

Give transformation clear ownership

An initiative can have an executive sponsor, a programme director and a full steering committee, yet still have nobody who can resolve its most consequential trade-off. The difficulty becomes visible when an enterprise priority conflicts with a business unit’s targets.

A new customer journey requires experienced people to leave their day jobs and help redesign the service. A technology investment creates benefits in several functions, but the cost sits in one. An AI-enabled workflow promises faster decisions, while introducing risks that no single department can assess alone.

Who can release the people? Who can move the funding? Who can accept the temporary reduction in performance while the organisation learns a different way of working? If those answers are unclear, the governance model is incomplete.

In Part 1, I argued that transformation must become a permanent leadership capability. That capability depends on an organisation being able to make and execute consequential decisions repeatedly.

Clear ownership makes that possible. It gives people a practical understanding of what they are responsible for, which decisions they can make and where they can obtain a decision when their authority ends. For senior leaders, the challenge is to make that clarity hold across organisational boundaries.

The difficult decisions sit between functions

Within a department, accountability can appear straightforward. There is a leader, a budget, a set of objectives and an established reporting line. Transformation exposes the limits of that arrangement because the outcome usually depends on several functions changing together.

Improving customer onboarding may require sales to collect information differently, operations to redesign case handling, technology to integrate data and risk to reconsider how controls are applied. HR may need to support new responsibilities and learning pathways. Finance must assess whether the investment produces a sustainable return. Every function can deliver its assigned activity while the overall customer experience remains disappointing.

The technology is deployed. Training is completed. The new process is documented. Yet customers still wait because exceptions move between teams without anyone having the authority to resolve them. This is the difference between assigning responsibility for individual activities and establishing ownership of the complete outcome.

An executive team should be able to name the person accountable for that outcome and explain the authority available to them. If the role depends entirely on persuasion, personal relationships and repeated escalation, its effectiveness will be fragile.

Deployment confidence does not settle operating-model questions

Deloitte’s 2026 Global Technology Leadership Study illustrates the tension. Among more than 660 technology executives surveyed, 81% said they could deploy and govern AI at scale. Yet nearly 75% acknowledged that their operating model would need to change within the next 12 to 18 months. [1]

These are executives’ assessments, rather than a direct measure of organisational readiness. They nevertheless raise an important leadership question. What does confidence in deployment mean if the organisation is still working out how decisions, funding and accountability need to change?

A system can be technically ready before the business is ready to take responsibility for how it operates. Consider an AI tool that recommends how customer cases should be handled. The technology team can assess system performance. Risk specialists can establish requirements for its use. Operations can train employees to review its recommendations.

But someone must still decide whether the complete workflow is ready for deployment, whether the remaining limitations are acceptable and whether the organisation has sufficient capacity to handle exceptions. Those decisions need to be designed into the governance model before the pressure of implementation arrives.

Give the executive team a clear decision-making role

My proposed model has two connected levels of governance, with delivery accountability embedded in the business. At the strategic level, a CEO-led executive group sets the outcomes, investment priorities and boundaries within which transformation will proceed. Its most valuable work concerns choices that individual functions cannot resolve.

Which opportunities deserve enterprise investment? Which existing commitments should be reduced? Where should the organisation accept temporary disruption? Which standards must remain consistent across the business? When should an initiative stop?

The CEO provides the authority to resolve enterprise trade-offs. The Chief Transformation Officer, where that role exists, coordinates the portfolio and brings dependencies into view. The CHRO helps connect organisation design, workforce capability and the experience of change. Finance, operations, technology and risk leaders contribute the decisions within their respective mandates.

Participation can be shared. Accountability for each material decision still needs to be specific. A committee should therefore be able to explain which decisions it makes collectively, which belong to a named executive and which have been delegated. It should also establish how disagreements are resolved.

Without that clarity, a steering committee can become a place where problems are described repeatedly without reaching a decision. The executive group must be willing to change the conditions of delivery. Asking a team to accelerate while leaving its funding, capacity and dependencies unchanged does not resolve the constraint.

Build a transformation office that helps decisions happen

At the execution level, a cross-functional transformation office connects enterprise architecture, organisation design, AI and data product management, finance and change enablement. Its role is to make the programme of change understandable and manageable across the enterprise.

That means identifying where initiatives compete for the same people, where one delivery depends on another and where a local decision could create problems elsewhere. It also means preparing decisions well enough that leaders can act. A useful escalation should explain the decision required, the available options, the consequences of delay and the recommendation. It should identify the person authorised to decide.

Reporting supports this work, but reporting volume is a poor measure of effectiveness. I would assess a transformation office partly through the obstacles it helps remove, the quality of the decisions it enables and the accuracy of its view of dependencies. It must also avoid becoming an additional approval layer for routine work. Teams need freedom to act within agreed boundaries. Central coordination is most useful where interdependencies, enterprise standards or material trade-offs require it.

As transformation becomes a permanent capability, these responsibilities should become part of the organisation’s normal management practices. Their value should survive changes in programme names and personnel.

Keep business leaders accountable for the transition

The business cannot outsource ownership of its future operating model to a central office. Business leaders understand the customers, operational constraints and practical consequences of change. They must help design the model and remain accountable for its performance after implementation.

That ownership should include the transition itself. A leader responsible for introducing a new workflow needs to know how service will be maintained, how employees will become proficient and how problems will be identified and corrected. Their responsibility cannot begin at launch and end with adoption figures.

The people doing the work must also contribute to the design. They often know where formal processes depend on informal judgement, workarounds or relationships that are absent from the documentation.

When they challenge a proposal, leaders should examine the evidence. A concern about workload may reveal an unrealistic transition plan. A warning about customer exceptions may expose a gap in the technology. Reluctance to remove a manual check may reflect experience of failures that the design team has not considered.

Some objections will need to be challenged. Others will improve the transformation. Treating disagreement as useful information makes it easier to distinguish between the two.

Define ownership through actual decisions

Consider an illustrative customer-onboarding redesign. The organisation wants to reduce delays while maintaining the quality of its risk decisions. AI will help extract information and identify inconsistencies. Employees will investigate uncertain cases and explain decisions to customers.

Naming an executive sponsor is a starting point. The organisation also needs to specify:

• Who owns the end-to-end customer outcome and the benefits expected from the investment?
• Who can change the workflow when evidence shows that it is underperforming?
• Who is authorised to approve deployment, subject to the required technical and risk assessments?
• Who can suspend the AI component when its performance falls outside agreed limits?
• Who resolves a conflict between a business target and a control requirement?

These responsibilities need to fit together. The outcome owner should have authority to coordinate delivery and a clear route for resolving resource conflicts. That authority should coexist with independent risk oversight and explicit limits on what can be approved.

Clarity also requires evidence thresholds. A team needs to know what must be demonstrated before a bounded experiment becomes a wider deployment. The requirements may include performance under realistic conditions, effective human review, operational support and a workable response to failure. The aim is to reduce uncertainty about how decisions will be made.

AI makes human accountability more consequential

Bain’s operating-model analysis places increasing emphasis on outcome ownership as AI takes on more execution. It also highlights the need to provide direction, guardrails and context as decision-making becomes more distributed. [2]

Delegating a task to a system does not settle who is answerable for its result. If an AI-supported process produces a poor outcome, the organisation needs to understand who owned the workflow, who approved its use and who was responsible for monitoring performance. Employees need to know when they should challenge an output and whether they have the authority to stop or escalate the work.

A requirement for “human oversight” is incomplete if the person reviewing the output lacks time, expertise or access to the underlying information. Leadership must make oversight workable. That includes allocating capacity for review, defining escalation routes and ensuring that raising a concern does not require someone to overcome pressure to meet an incompatible target.

Align incentives with the outcomes you expect

Governance will struggle if the organisation rewards behaviour that works against it. A business leader measured solely on departmental cost may resist releasing experienced employees to an enterprise initiative. A sales leader rewarded only for volume may pass incomplete cases downstream. A technology team judged by deployment dates may have little incentive to examine whether the new process improves customer outcomes.

These responses can be understandable within the management system leaders have created. Shared outcomes should therefore appear in performance expectations, funding decisions and benefit reviews. Leaders need visibility of costs and gains across the whole workflow, including burdens transferred between teams.

Be explicit about who funds the transition, who provides scarce expertise and how temporary disruption will be treated in performance discussions. Accountability is more credible when people have both the authority and the resources to meet it.

Use governance to shorten the path to a sound decision

Speed depends partly on how long the organisation spends waiting for decisions. Review cadence should reflect the decision involved. A routine dependency may be resolved through a short fortnightly discussion. A material incident may require immediate action. A major investment decision may need deeper analysis and formal challenge.

More meetings will not automatically improve any of these. At the next steering meeting, take one blocked decision and trace its path. Establish when it first became visible, what information was missing, who had the authority to decide and why the decision remained unresolved.

Then change the arrangement that allowed it to stall. Perhaps the team lacked delegated authority. Perhaps two leaders held conflicting mandates. Perhaps the decision was repeatedly discussed in a forum that could not approve it.

That examination will tell you more about the effectiveness of governance than another layer of status reporting. Transformation ownership becomes real when leaders can make difficult choices and the organisation can act on them. The test is whether a person facing a consequential trade-off knows what they can decide, what they must escalate and who will take responsibility for the answer.

Where does accountability become unclear when transformation crosses the boundaries of your organisation—and what decision will you make to clarify it?

Next in the series: Redesign the work before redrawing the organisation.

Research underpinning this article:

[1] Deloitte Insights (29 June 2026), Rewiring the Enterprise Operating Model for AI Scale.

[2] Bain & Company (8 May 2026), An Operating Model for the Age of AI.

The onboarding example is illustrative. The proposed governance model and leadership practices are a synthesis informed by the research.