.png)
We’ve invested heavily in AI but aren’t seeing enough business value. What kind of leadership do we actually need?
We’ve invested heavily in AI but aren’t seeing enough business value. What kind of leadership do we actually need?
Many organisations are no longer asking whether to invest in artificial intelligence. They have already committed budget, built teams, run pilots and introduced AI into parts of the business.
They're now asking a harder question.
Why is the investment not producing enough measurable business value?
It is reasonable to ask whether the organisation needs stronger or different AI leadership, including whether a Chief AI Officer would provide clearer accountability.
Sometimes that is the right answer. But only sometimes.
Poor returns from AI can reflect unclear business priorities, fragmented ownership, weak technology and data foundations, cross-functional friction or governance that slows rather than enables decisions. They can also reflect leadership capability that is no longer sufficient for the next stage.
The leadership you need depends on where value is being lost.
The gap between AI investment and business value is real
Lucent Search's research with more than 100 AI and data leaders found significant friction between ambition and execution.
85%
reported difficulty measuring the return on AI investment
91%
struggle to move AI pilots into production
87%
report challenges with cross-functional collaboration
71%
struggle to secure sufficient budgets for AI initiatives
These problems are often joined up.
AI leaders who cannot demonstrate value find it harder to secure further investment. Pilots that cannot integrate into existing technology or operating processes remain experiments. Business functions that are not accountable for adoption can leave specialist AI teams trying to create change through influence alone.
Often, the issue is not necessarily a shortage of AI expertise. The issue is whether the organisation has created the conditions for success.
McKinsey reported in June 2026 that almost 90% of organisations were experimenting with AI, but only 7% said they had scaled it across the enterprise.
Investment and activity are therefore poor proxies for execution maturity.
Before changing the leadership team, establish what is actually failing
Six different problems can sit behind disappointing AI returns. It is not uncommon for several to exist at the same time.
01 — Value
Is the organisation clear about where AI should create business value?
An organisation can have considerable AI activity without a sufficiently disciplined view of what that activity is meant to achieve.
A long list of use cases is not a strategy.
Is somebody making clear choices about where AI can materially change economics, customer outcomes, operational performance or risk? Do they know where AI cannot do this?
If priorities are unclear, appointing another technology executive may simply give somebody accountability for an unfocused portfolio, or worse, they may create a portfolio full of nice-to-have tools that don’t move the needle.
Your first requirement may instead be to set the value agenda. Fewer priorities and clearer outcomes can support more explicit decisions about where investment should concentrate.
The useful question is not:
What can we use AI for?
It is:
Which business outcomes are important enough for us to redesign work, allocate capital and hold an executive accountable for delivering them?
Business ownership matters because AI value is created in the business
Many AI programmes begin in technology, data or innovation functions.
That can be appropriate while capability is being established. However, it becomes problematic when responsibility for business results stays there.
A technology team can build a forecasting model. It cannot, on its own, make another part of the business change how it works.
A data function can develop customer intelligence. It cannot independently change commercial behaviour.
An AI team can automate part of a process. It does not automatically own the productivity, customer or financial outcome that justified the investment.
BCG's 2026 research makes a similar distinction: the CEO needs to orchestrate the AI agenda, while delivery accountability sits with executives and P&L owners responsible for the underlying business outcomes.
That suggests a useful test:
If an AI initiative succeeds technically but produces no business result, which executive is accountable?
If you don’t know the answer, the organisation has an ownership problem.
02 — Ownership
Does a business executive own the result, or merely support the initiative?
The strongest structure usually separates capability from outcome.
Technology, data and AI leaders may provide platforms, architecture, specialist capability, governance, standards and technical expertise.
However, the business still needs to own what changes.
That does not diminish the role of senior AI leadership. It makes how their success is measured clearer.
When business leaders see AI as something being delivered to them by a central function, adoption can remain optional. When they own the operational, commercial or customer outcome, the conversation changes.
The question becomes less about whether the technology works and more about what the organisation must do differently to realise the value.
Sometimes the constraint really is technology and data
Lucent Search's research found that 91% of AI leaders struggle to move pilots into production, while 89% face challenges integrating AI with legacy systems.
That matters because an impressive pilot is very different from a production capability.
Enterprise AI has to operate through real systems, real data, real processes and real controls.
If an organisation repeatedly demonstrates technically credible AI but cannot integrate it into the environments where decisions and transactions occur, the leadership problem may sit primarily with technology architecture and engineering.
That points towards the CIO or CTO rather than automatically towards a dedicated AI executive.
Equally, if progress is being constrained by fragmented data, unclear data ownership, poor quality or insufficient analytics capability, the CDO or CDAO may have the more important role to play.
03 — Foundations
Is the organisation unable to create value because AI cannot reliably connect to the technology and data on which the business depends?
If so, hiring a senior executive whose principal strength is AI strategy may not address the bottleneck.
The organisation may need a CIO capable of modernising enterprise technology, a CTO able to strengthen engineering and architecture, a CDO or CDAO able to establish usable enterprise data — or a clearer view of priorities for one of the executives already in place.
The underlying question is:
What capability has to change before AI can operate at enterprise scale?
That question should be answered before deciding which title to recruit.
Cross-functional friction becomes a leadership issue when nobody can resolve it
AI rarely sits neatly inside one function.
Delivery can require technology, data, legal, risk, finance, operations, HR and individual business units to make connected decisions.
Lucent Search's research found that 51% of AI leaders described cross-departmental collaboration as very challenging and a further 36% described it as moderately challenging.
This creates a different kind of leadership problem.
The organisation may have excellent technical leaders and credible business sponsorship, yet still lack somebody with enough authority to resolve competing priorities across the system.
That is where enterprise AI leadership can become valuable.
04 — Authority
Does somebody have sufficient authority to make the organisation work across its existing boundaries?
This is one of the stronger arguments for a dedicated Chief AI Officer.
Not because AI automatically deserves another C-suite title, but because an enterprise-wide agenda may require somebody who can coordinate decisions across technology, data, governance, investment and business adoption.
But the title only works if the authority follows it.
An executive expected to transform the organisation through AI while controlling little budget, owning few resources and depending on informal influence for every important decision has not been given an enterprise remit.
They have been given an enterprise expectation.
Budget can reveal whether the role has real authority
71%of AI leaders in our research reported difficulty securing sufficient budgets for AI initiatives.
Budget pressure is not inherently evidence of poor leadership. Access to capital should be challenged, and weak use cases should not be funded indefinitely.
The more important question is how investment decisions are made.
Does the executive responsible for AI have a credible route to capital when evidence justifies scaling?
Can funding move from experimentation into production?
Can investment be redirected away from low-value initiatives?
Can someone connect expenditure to business outcomes in terms the CEO, CFO and board can evaluate?
Where that mechanism is missing, accountability becomes disconnected from resources.
A leader cannot reasonably own an enterprise outcome while repeatedly negotiating for the basic authority, headcount and investment required to pursue it.
Governance should enable decisions, not simply create controls
AI creates legitimate questions around risk, privacy, regulation, model performance, third-party providers and responsible use.
Governance therefore matters.
But organisations can respond to uncertainty by adding layers of approval without clarifying who is empowered to decide.
The result can be technically compliant experimentation that never becomes operational capability.
The leadership requirement is not simply someone who understands AI governance.
It is leaders who can make disciplined trade-offs between value, speed and risk. To do this, they need to know which decisions belong with technology, data, legal, risk, business leadership or the board.
05 — Decision rights
Are ownership, escalation and decision rights clear enough for the organisation to move?
If every significant decision requires negotiation between several executives, the problem may be structural.
A new hire will not fix that unless their mandate changes the decision architecture around them.
The organisation needs to be explicit about who can approve investment, who owns risk, who can stop deployment, which decisions should sit with business leaders and which issues genuinely require executive committee or board involvement.
Good governance does not mean centralising every decision.
It means making clear where decisions belong and giving the people responsible enough authority to make them.
Sometimes the structure is not the problem
There is a risk in over-diagnosing organisational design.
Not every stalled AI programme is the result of unclear responsibilities, fragmented authority or an immature operating model.
Sometimes the organisation has made sensible choices about priorities. Business ownership is reasonably clear. Technology and data foundations are good enough. Projects are well funded.
If leadership is still not delivering, this needs to be considered directly.
06 — Capability
Does the current leadership team have the capability required for the next phase?
Moving from AI experimentation to enterprise value can require a different combination of leadership capabilities from those needed to establish an initial programme.
The organisation may now need leaders who can connect AI investment to material business outcomes, make informed trade-offs across technology, data, risk and commercial priorities, move capability into production, change workflows and operating models, and build confidence with the executive team and board.
These capabilities cannot be inferred from technical knowledge alone.
A leader may understand AI extremely well but struggle to convert it into decisions the wider organisation can act on.
Another may be an accomplished enterprise executive but lack enough understanding of AI, data or engineering to challenge assumptions and judge where investment should go.
The relevant question is not:
Does this executive understand AI?
It is:
Do they have evidence of leading the kind of organisational change now required?
This distinction matters because a structural problem and a capability problem require different interventions.
If the leader is capable but constrained by unclear authority, replacing them may reproduce the same failure with somebody new.
If the role responsibilities are coherent but the leader cannot deliver against them, changing reporting lines will not solve the problem either.
Before making an appointment, establish which of those two situations you are dealing with.
So what kind of leadership do you actually need?
These six questions separate problems that can be addressed through clearer priorities, ownership or operating model from those that genuinely require different leadership capability.
Only then is it useful to decide what kind of leader the organisation needs.
If the problem is unclear business value
The organisation may need stronger business leadership, not another technology title.
The priority is to connect AI investment to a smaller number of material business outcomes, make explicit choices about where investment should go and establish accountable business owners.
If the problem is technology integration and production
The CIO or CTO may play a more important role.
The role should centre on enterprise architecture, engineering, platforms, integration, resilience and turning technical capability into production systems.
The distinction between CIO and CTO will depend on whether the central issue is enterprise technology and operating systems, or engineering, architecture and technology-enabled products.
If the problem is data
The organisation may need stronger CDO or CDAO leadership.
The mandate may include enterprise data ownership, governance, analytics, platforms and the data foundations on which AI depends.
Adding AI to an existing data remit only works if the executive has sufficient authority and resources to deliver it.
If the problem is enterprise coordination
A Chief AI Officer can make sense when AI genuinely requires distinct enterprise accountability across technology, data, risk, investment and the business.
That case becomes weaker if an existing CIO, CTO or CDAO already has sufficient scope and authority to own the agenda.
The test is not how important AI is.
It is whether an enterprise responsibility is currently both material and genuinely unowned.
If the problem is adoption inside individual functions
The answer may be stronger business ownership, supported by central AI, data and technology capability.
Creating a central executive to compensate for weak accountability in the business can move ownership further away from where value is supposed to occur.
If the problem is leadership capability
The organisation may need a different executive rather than another executive.
Before creating a new C-suite role, define what the current leadership cannot deliver and whether that capability should sit within an existing CIO, CTO, CDAO or AI leader.
This is where external market calibration becomes useful.
A leadership requirement that sounds sensible internally may combine responsibilities that rarely coexist in one person, require authority the role will not actually possess, or demand experience the organisation cannot realistically attract at the proposed level.
Test the role with an executive search specialist before the search begins.
When a new executive appointment is justified
A new appointment becomes more credible when the organisation can identify a leadership requirement that is both important and genuinely unowned or underpowered.
For example, when:
nobody has enterprise accountability for AI outcomes;
an existing CIO, CTO or CDAO remit cannot realistically absorb the agenda;
the organisation needs materially different capability for the next phase;
current executives lack the experience or authority required to move from experimentation into scaled operations; or
the existing leadership architecture contains a gap that cannot be resolved through clearer decision rights alone.
The mandate should then be defined around what needs to change rather than around whichever executive title is currently attracting attention.
That may produce a Chief AI Officer role.
It may produce a different CIO, CTO, CDAO or COO mandate.
It may reveal that no additional C-suite role is required.
That is a useful outcome too.
The first leadership decision is not who to hire
When significant AI investment is failing to produce enough value, boards understandably want intervention.
But creating a new executive position before diagnosing the execution problem risks adding another layer to an already fragmented system.
Start instead with where value is being lost.
Is the organisation pursuing the right outcomes?
Does the business own them?
Can AI reach production through the existing technology and data environment?
Does somebody have enough authority to resolve cross-functional dependencies?
Are governance and decision rights helping good initiatives move responsibly?
And does the current leadership team genuinely have the capability required for the next stage?
Only then should the organisation decide whether it needs a new role, a different executive or a stronger mandate for somebody already in place.
The first leadership decision is not who to hire. It is what needs to be owned, by whom, with what authority. Then whether the people currently responsible are capable of delivering it.
Related analysis and executive search
Do you need a Chief AI Officer — or should AI sit with the CIO, CTO or CDAO?
A framework for deciding where enterprise AI accountability should sit.
For organisations creating or redefining enterprise AI leadership.
For organisations where data, analytics and AI leadership need clearer ownership.
For organisations where enterprise technology, integration and transformation are central to the challenge.
Is the AI leadership problem clear enough to hire against?
Lucent Search works with boards and executive teams to establish what needs to change, where accountability should sit and what leadership capability is required before going to market.
Where a new appointment is the right answer, Lucent conducts the retained executive search.
Comments