ADVISORY AREA / 03
Leadership &
Enterprise Performance
A board approves the strategy., the executive team agrees the priorities, capital is allocated, targets are clear.
Six months later, the organisation is moving, but not in the way anybody expected.
Decisions travel upwards that were supposed to be made locally. Senior executives spend increasing amounts of time resolving matters that should never have reached them. One division performs strongly while another uses almost the same operating model and repeatedly misses plan.
A transformation programme is technically on schedule and commercially behind. The chief executive is involved in so many decisions that their involvement has become a control mechanism.
Management responds by adding meetings, reporting, governance and programme management. The organisation becomes more supervised, not necessarily more capable.
This is where Leadership & Enterprise Performance work at E. Cartier & Co begins. Not with the assumption that the people are poor. Not with another leadership competency exercise. And not with the idea that better execution simply requires more accountability.
We examine something more fundamental:
The organisation can have the right strategy and still be
structurally incapable of delivering it.
The useful question is not simply how much growth the plan produces. It is what the enterprise becomes while producing it.
Because execution rarely fails in one place. It is usually distributed across authority, information, incentives, capability, operating rhythm and leadership behavior.
And increasingly, another participant has entered the system: technology is beginning to make decisions too.
When hundreds of people and autonomous systems can act on behalf of the enterprise, who actually owns the result?
01
Authority
Who has the right to decide, and where does that authority stop? A job title is not a decision architecture.
02
Accountability
Can responsibility for an outcome be traced to somebody with enough authority, information and capability to influence it?
03
Execution & Adaption
Does the organisation convert strategic priorities into coordinated operating behavior, or merely into more programmes?
When assumptions change, can the enterprise change course without waiting for the annual strategy cycle or the chief executive to intervene?
01 / THE ECONOMICS OF ESCALATION
A surprising amount of poor performance begins with a reasonable management decision.
An executive team wants more control after a difficult year.
More reporting is introduced.
Approval thresholds are tightened.
The centre becomes more involved.
Nothing about that decision is obviously unreasonable.
Then something changes.
Divisional leaders learn that taking a difficult decision independently carries more personal risk than escalating it.
Investment decisions move upwards.
Commercial exceptions move upwards.
Hiring decisions move upwards.
Customer disputes move upwards.
Within months, the executive committee is discussing matters that would never have appeared there two years earlier.
Management concludes that the divisions lack accountability.
The divisions conclude that headquarters does not trust them.
Both observations contain some truth.
Neither identifies the underlying problem.
The organisation has changed the economics of decision making. Escalation has become safer than judgement.
That is not primarily a cultural issue.
It is an operating system issue.
02 / COMPETENT PEOPLE, WEAK SYSTEM
Performance can deteriorate while every executive appears individually competent.
This is one of the more uncomfortable realities of enterprise performance.
An organisation can contain intelligent, experienced and hardworking leaders and still execute badly.
The chief financial officer may be right to insist on tighter capital discipline.
The chief commercial officer may be right to push for faster customer decisions.
The chief operating officer may be right that standardisation is necessary.
Divisional managing directors may be right that local markets require flexibility.
The chief technology officer may be right that legacy systems are preventing scale.
The CEO may be right that too many initiatives are moving too slowly.
No individual has to be obviously wrong for the system to become dysfunctional.
The problem appears in the interfaces.
Finance slows a decision commercial considers urgent.
Commercial sells a proposition operations cannot deliver economically.
Operations creates local workarounds technology is trying to remove.
Technology implements standardisation that one division believes destroys an important competitive advantage.
Head office asks for accountability while retaining the right to override decisions.
The organisation experiences these as separate problems.
They may actually be one problem:
Important decisions are crossing organisational boundaries without a sufficiently clear architecture for who decides, who contributes, who can stop the decision and who owns the result afterwards.
That is where performance begins leaking.
03 / EXECUTIVE BOTTLENECK
The executive team can become the organisation's most expensive bottleneck.
Some chief executives are proud that they remain close to everything.
At a certain scale, that can become a warning.
A CEO who approves the largest contracts, reviews senior hires, intervenes in customer disputes, signs off investment exceptions and resolves disagreements between divisions may be extraordinarily capable.
The company may also have built an operating model around that capability.
This often works for longer than expected, particularly in founder led organisations.
The founder knows the customers.
They understand the commercial history.
They can identify weak numbers quickly.
They know which executive can actually deliver.
A fifteen minute conversation replaces a process that would otherwise take a week.
Then the company doubles.
Or acquires three businesses.
Or expands internationally.
Or the founder wants to spend more time on strategy and less on daily operations.
Suddenly, the mechanism that made the organisation fast becomes the mechanism preventing it from scaling.
The answer is not simply “delegate more”.
Delegation without the information, capability, decision boundaries and escalation rules to support it can create a different failure.
Leadership has to become institutional without making the institution inert.
That is a harder design problem.
04 / AUTHORITY BEFORE ACCOUNTABILITY
Accountability is frequently demanded where authority has never been given.
A divisional managing director is responsible for EBITDA.
Headquarters controls pricing exceptions.
Capital expenditure above a relatively low threshold requires group approval.
Senior hires require another approval.
Procurement has been centralised.
Technology priorities are decided elsewhere.
A large customer account crosses three divisions.
The managing director still owns the number.
Management then asks why they are not behaving like an owner.
Sometimes the answer is uncomfortable.
The organisation has given somebody accountability for an outcome while retaining many of the decisions that determine it.
The reverse problem also exists.
A leader has significant autonomy.
Performance deteriorates.
Head office intervenes only after the damage is visible.
Nobody can say which indicators should have triggered earlier intervention, because autonomy was never accompanied by an explicit standard for when it should end.
The organisation oscillates between freedom and control.
Neither is properly designed.
For each material outcome, does the accountable person possess the authority, information, resources and intervention rights required to influence it?
If not, accountability exists largely on paper.
05 / WHAT THE BOARD CAN AND CANNOT SEE
The board sees the organisation through a very narrow window.
Most boards receive large amounts of information.
That does not mean they have an accurate picture of execution.
Board packs tend to be strongest where the organisation is already good at measurement.
01
Revenue
Visible, comparable and usually already historical by the time the board sees it.
02
Margin
Important, but often too late to explain the operating choices that produced it.
03
Cash
Consequential, yet rarely tells the board where execution is becoming fragile.
04
Headcount
Easy to count, difficult to interpret without knowing capability and dependency.
05
Customer metrics
Useful, but can hide unresolved commercial and operating trade offs.
06
Programme milestones
Often describe delivery activity rather than realised enterprise value.
Those are important.
They are usually lagging indicators of something that has already happened.
The more difficult questions often sit between the numbers.
Which executive decision has been deferred for three months because nobody owns it?
Where is the same operating problem being solved differently in four divisions?
Which transformation programme is reporting green because the implementation milestones have been achieved while the commercial benefit has not?
Which senior leader has become indispensable to a process that was supposed to be institutional?
Where are people routinely bypassing the formal organisation because the informal route works better?
How many decisions require the CEO because management has not agreed what happens when two functions disagree?
These do not fit neatly onto a dashboard.
They can still determine whether the strategy works.
06 / ACTIVITY IS NOT VALUE
Execution is not the number of initiatives marked green.
There is a particular danger in large transformation portfolios.
An organisation approves twelve programmes.
Each has a sponsor.
Each has a project team.
Each has milestones.
A transformation office consolidates the reporting.
The board receives a page showing that nine are green, two amber and one red, It feels controlled.
Then the financial year ends.
The benefits do not resemble the original business cases.
Everybody can explain why.
Implementation was completed, but adoption was weaker than expected.
The cost saving required behavioural change that never occurred.
Revenue benefits relied on commercial teams using the new system differently.
Technology was delivered but the operating model remained largely unchanged.
Several benefits were counted by more than one programme.
None of this necessarily implies incompetence.
It means programme completion and enterprise performance were being treated as though they were the same thing.
They are not.
A transformation programme is useful only to the extent that something materially different happens in the enterprise afterwards.
That distinction sounds obvious.
In practice, organisations lose it remarkably easily.
07 / AUTONOMOUS WORK
Artificial intelligence makes an old governance problem much harder.
For most of corporate history, the organisation could assume that consequential decisions were made by identifiable human beings.
That assumption is weakening.
An AI system prioritises leads.
Another recommends pricing.
An agent drafts supplier negotiations.
A customer service system resolves complaints.
A procurement tool changes reorder quantities.
A finance model flags transactions for intervention.
A recruitment system determines which applicants receive human attention.
Individually, these applications can look relatively modest.
Collectively, the enterprise begins delegating thousands of judgements.
Now consider the governance question.
Who owns the result when the system acts correctly according to its objective and the enterprise dislikes the outcome?
The technology team?
The executive who approved deployment?
The business owner?
The employee who accepted the recommendation?
The vendor?
The model?
None of those answers is satisfactory on its own.
The more autonomous work becomes, the more important decision architecture becomes.
Authority needs boundaries.
Human intervention needs defined triggers.
Exceptions need owners.
Performance needs to be attributable.
Some decisions should never be delegated.
Others probably should have been delegated years ago.
The question is not whether AI belongs in the organisation. It is whether the organisation is capable of governing decisions that no longer originate entirely from people.
08 / AN ENTERPRISE DECISION
£860m
Annual turnover
5
Operating divisions
3
Acquisitions completed over
four years
£1.2bn
Board growth ambition
7 years
CEO tenure
Growth has worked commercially. The operating model is beginning to tell a different story.
Consider a European industrial services and infrastructure group producing £860 million in annual revenue.
The group has grown quickly, some of that growth was organic.
A significant proportion came from three acquisitions completed over four years.
The businesses are profitable.
Customers generally regard the group well.
The acquisition thesis has worked commercially.
Operationally, the company is beginning to feel different.
There are now five operating divisions.
Two retain much of their original management.
One has been integrated heavily.
Another was deliberately left autonomous because performance was strong.
The fifth contains a mixture of legacy and acquired operations.
The board has approved a strategy to move beyond £1.2 billion of revenue while improving margins.
The strategy depends on several things happening together.
Cross selling across divisions.
Better utilisation of technical staff.
Greater procurement leverage.
A common data environment.
Automation of administrative work.
More consistent commercial discipline.
And another potential acquisition.
None of these is individually unreasonable.
The executive committee is experienced.
The CEO has been in post for seven years and understands the company unusually well.
There is no obvious leadership crisis.
Yet performance is becoming harder to predict.
The first symptom looks like a margin problem.
One division misses its margin target.
The explanation is credible.
Labour utilisation weakened.
A large contract mobilised later than expected.
Procurement savings arrived more slowly.
The board accepts the explanation but expects improvement.
Three months later, another division reports a similar utilisation issue, the causes appear different.
Management asks the chief operating officer to conduct a group review.
That review identifies something more interesting.
Each division defines operational capacity differently.
One counts subcontractors inside the utilisation measure.
Another does not.
One division books internal technical support as project time.
Another records it centrally.
A third has developed its own scheduling process because the group system is considered too slow.
The headline metric is called the same thing, operationally, it means different things.
The company does not have a performance problem yet. It has a problem knowing precisely which performance problem it has.
Then the CEO's diary tells a second story.
During the same review, somebody looks at the chief executive's involvement in significant operating decisions.
The result is uncomfortable.
A major customer contract required the CEO to resolve disagreement between commercial and operations.
Two capital requests were escalated because finance and divisional management could not agree how the return should be calculated.
A senior appointment waited eleven days because two executives disagreed about whether the role belonged to the group or the division.
An acquisition integration issue reached the CEO after neither the group function nor acquired management believed they held sufficient authority to decide it.
The CEO dealt with all of them effectively.
That was the problem.
Nothing had failed dramatically enough to force redesign.
The organisation had learned that difficult ambiguity could be sent upwards and resolved by somebody unusually capable.
The CEO was not merely leading the operating system.
They had become part of it.
The executive committee is busy because the organisation has not decided where authority belongs.
The board's initial response could easily be to ask management to delegate more.
That would be premature.
A closer look shows several different problems.
Some decisions genuinely belong at group level.
Others should never have reached it.
Certain decisions require collaboration between functions but have no agreed final decision maker.
The acquired businesses still operate under assumptions that pre date the group's current structure.
Several senior executives use influence to solve problems where formal authority is unclear.
This has created an organisation that works surprisingly well when the right people are present.
It becomes much less reliable when they are not.
That distinction matters.
Strong relationships between executives are valuable.
They should not be the only mechanism preventing structural ambiguity from becoming commercial failure.
The proposed AI programme exposes the problem more clearly.
The group intends to deploy automation and AI across scheduling, procurement, commercial analytics and administrative processes.
The business case is substantial.
One pilot already shows meaningful potential.
The technology team wants to accelerate.
Then somebody asks a basic question. The scheduling system will be allowed to recommend reallocating skilled employees between projects.
Who can override it?
The divisional managing director?
The group operations team?
The project leader?
What happens where moving somebody improves group utilisation and damages the economics of the division that currently employs them?
The procurement agent may negotiate within pre approved parameters.
Who owns the supplier relationship when it does something commercially rational but strategically unhelpful?
The commercial system will identify price opportunities.
Can a salesperson ignore it?
At what level?
How frequently before somebody intervenes?
The AI programme has not created these questions.
It has exposed decisions the human organisation was already handling through judgement, relationships and escalation.
Automation simply removes the ability to pretend the ambiguity does not exist.
This becomes an enterprise design issue rather than a technology implementation.
The performance architecture begins with decisions, not the organisation chart.
The company could redraw reporting lines.
It might help, but organisation charts can create the appearance of precision while leaving the actual decision system largely unchanged.
Instead, the executive team starts with a relatively small number of decisions that materially affect performance.
Who controls pricing outside standard parameters?
Who decides whether a major customer requirement justifies operational exception?
Who allocates scarce technical capacity between divisions?
Who can approve capital?
Who owns group customer relationships?
Who determines whether an acquired business retains a local process?
Who decides when group standardisation overrides divisional advantage?
Who can stop an AI supported decision?
For each one, the organisation distinguishes several things.
Who recommends.
Who contributes.
Who decides.
Who executes.
Who can challenge.
Who is accountable for the result.
And what evidence should trigger escalation.
The answers do not all move towards headquarters.
Some decisions move the other way.
That is important.
The purpose is not centralisation, it is to stop ambiguity being resolved accidentally.
More autonomy initially makes some performance worse.
This is the sort of detail that matters.
Suppose several commercial decisions move back to divisional leaders, they now have greater pricing freedom within clearer boundaries.
One division uses that freedom well.
Another discounts more aggressively than management expected.
For two quarters, the result is uncomfortable.
The old system would probably have prevented several of those decisions.
It would also have preserved the dependency on group approval.
The executive team now has a choice.
Take the authority back, or use the mistakes to improve the decision standard.
That is what institutionalising judgement actually looks like.
The objective is not to design an organisation in which nobody makes the wrong call, such an organisation would probably make very few calls at all.
The aim is to make responsibility clearer, detect poor decisions earlier and improve the quality of judgement without forcing every important decision back to the centre.
Performance truth becomes an enterprise capability.
The margin review also changes.
Instead of immediately imposing another group dashboard, management first agrees which measures have to mean the same thing everywhere.
Not every metric does.
Some businesses genuinely operate differently.
The distinction is made deliberately.
Utilisation receives one group definition where management needs comparability.
Divisions retain additional local measures where those are commercially useful.
Transformation programmes are no longer assessed only on delivery milestones.
The benefits have owners outside the programme team.
A technology implementation can therefore be technically complete and commercially unfinished.
That sounds like a small reporting change, It is not; it changes who has to explain the result.
The organisation begins separating activity, completion and value realisation. Those are three different states.
Leadership capacity is treated as finite.
The strategy still contains substantial ambition, but management stops pretending executive attention is infinitely divisible.
The board identifies the handful of programmes requiring genuine enterprise leadership.
Other initiatives are delegated further down.
Some are sequenced.
One is stopped.
Not because its business case was poor, because the organisation could not execute everything simultaneously to the standard required.
The chief executive's role changes too.
They remain involved in consequential decisions.
But the pattern becomes more deliberate.
Intervention is expected where a defined threshold is crossed, not simply because somebody believes the CEO will produce the fastest answer.
That creates space, not only in the diary bit in the organisation.
A company becomes more scalable when senior judgement can be reproduced through standards, capable people and clear authority rather than through permanent access to the same individual.
AI governance becomes part of operating governance.
The technology programme also proceeds, but it is no longer governed as though AI were a separate category of enterprise activity.
For each meaningful automated decision, management determines:
what the system is permitted to decide;
what remains a recommendation;
what information it can use;
which outcomes require human review;
who owns performance;
how exceptions are logged;
what happens when the model behaves within specification and still produces an undesirable commercial outcome;
and when authority should be withdrawn.
Some use cases accelerate, others are narrowed.
One is paused, not because the technology is weak.
Because the organisation cannot yet define the accountability around it confidently enough.
That is a legitimate reason not to automate a decision.
The return on AI does not come from the volume of activity delegated to machines. It comes from decisions the enterprise can delegate without losing control of consequence.
There is no perfect ending.
The redesigned organisation will still have politics.
Executives will still disagree.
A divisional leader may stretch the limits of their authority.
A group function may try to reclaim control after a local mistake.
The CEO will occasionally intervene where the formal process suggests they should not.
An AI system will eventually produce an output nobody anticipated.
Some performance measures will remain imperfect.
The point is not to create an organisation without friction.
Friction is often evidence that genuinely different interests are being reconciled.
The objective is to know which friction is productive and which is simply the cost of ambiguity.
That distinction allows leadership to intervene intelligently.
09 / WHAT E. CARTIER & CO ACTUALLY BUILDS
Leadership & Enterprise Performance work should leave behind an operating architecture, not merely a diagnosis.
Depending on the mandate, that can include:
A clear view of the consequential decisions that drive performance, including who recommends, decides, executes, challenges and owns the outcome. This is usually more useful than beginning with reporting lines.
Enterprise Decision Architecture
The boundaries of management authority and the specific conditions under which a decision should move elsewhere. Escalation becomes a designed mechanism rather than the safest response to uncertainty.
Authority & Escalation Map
A coherent view of the measures the organisation actually needs in order to understand performance. Definitions, ownership, leading indicators, lagging indicators, exceptions and intervention thresholds. Not simply more reporting.
Enterprise Performance System
The rhythm through which leadership reviews performance, resolves cross enterprise decisions and intervenes where necessary. Different conversations require different forums.
Executive Operating Cadence
A view of where a small number of people have become critical dependencies across the strategic portfolio. This allows the organisation to distinguish nominal accountability from actual executive load.
Leadership Capacity Map
The small number of initiatives capable of materially changing enterprise performance, considered together rather than as unrelated programmes. This makes sequencing, resource conflict and duplicated benefit assumptions visible.
Strategic Execution Portfolio
Decision boundaries, ownership, intervention standards and accountability for AI enabled and increasingly autonomous work. Technology governance is connected to enterprise governance rather than left as a parallel system.
AI & Autonomous Decision Governance
Clear principles for when the centre should intervene, when it should not and what evidence justifies overriding local authority. Autonomy without intervention standards eventually becomes either neglect or centralisation.
Performance Intervention Standard
The purpose of these assets is not to make the organisation more bureaucratic.
It is to reduce the amount of bureaucracy required to compensate for unclear decisions.
10 / WHERE THIS WORK BECOMES NECESSARY
The need is often visible before anybody describes it as an enterprise performance problem.
The chief executive is being pulled into too many operating decisions.
Acquisitions have created multiple ways of doing the same thing.
The organisation has grown faster than its management structure.
A transformation portfolio contains more activity than realised value.
Divisions are accountable for outcomes they do not fully control.
Head office has centralised decisions and performance has become slower, or headquarters has decentralised aggressively and cannot now see problems early enough.
Leadership succession is approaching and too much institutional knowledge remains concentrated in a few people.
AI and automation are scaling faster than the organisation's governance model.
Performance varies materially between business units without a convincing explanation.
The board receives increasing amounts of information and decreasing confidence about what is actually happening.
These situations do not always require restructuring.
Sometimes restructuring makes them worse.
They require diagnosis before organisation design.
11 / WHEN NOT TO REORGANISE
We would not recommend a reorganisation simply because the organisation is uncomfortable.
There is an understandable tendency to change structure when performance weakens.
A new reporting line is visible.
A new role feels decisive.
Boxes move on the organisation chart and management can point to action.
Sometimes that is exactly what is required.
Sometimes the structure is not the problem.
A division may simply have the wrong leader.
A commercial model may be weak.
An incentive may reward behaviour management says it dislikes.
The strategy may contain contradictory priorities.
A technology programme may be failing because the underlying process is poor.
Centralisation may have become excessive or the supposedly dysfunctional informal organisation may actually be compensating intelligently for defects in the formal one.
In those cases, restructuring can destroy something useful without solving the underlying issue.
We would also be cautious where a board wants “greater accountability” without being prepared to transfer authority, or wants greater autonomy while retaining the ability to override every material decision.
Those positions are internally inconsistent.
The enterprise has to choose more precisely.
12 / QUESTIONS WORTH ASKING
Questions worth asking when the organisation is working hard and still not moving properly.
Which decisions repeatedly reach the executive committee that should not require executive committee judgement?
The answer usually reveals where authority is unclear or capability is weak.
Who owns an outcome today without controlling enough of the decisions that determine it?
Accountability without authority creates frustration before it creates performance.
Which executive would cause the greatest operational disruption if they disappeared for ninety days?
That is an institutional dependency whether or not anybody has called it one.
Where does the organisation report programme completion without being able to demonstrate the corresponding enterprise benefit?
Activity, delivery and value realisation should not be confused.
Which important process works because the right people know one another rather than because the organisation has designed it to work?
Relationships are an asset. Hidden structural dependency is not.
What decision have we centralised because one person made a mistake?
The cost of preventing a repeat may be greater than the original mistake.
Which AI supported decision could materially affect a customer, employee, supplier or balance sheet without one identifiable executive owning the consequence?
That question will become harder to ignore.
If the chief executive stopped resolving ambiguity tomorrow, where would the organisation stop?
That is where the next stage of institutional capability has to be built.
CLOSING PERSPECTIVE
Execution is more complicated than “strategy decided, people execute, results follow”.
The word “execution” can make enterprise performance sound simpler than it is.
Strategy is decided.
People execute.
Results follow.
Real organisations are messier.
Priorities compete.
Information arrives late.
Strong executives disagree for good reasons.
A decision which looks obvious from the centre appears irrational in the market where it has to be implemented.
One business unit needs standardisation.
Another creates its advantage precisely by doing something differently.
A process is inefficient because it is protecting the organisation from a risk nobody remembered to document.
The CEO intervenes because the formal system is too slow.
Then the formal system never improves because the CEO keeps intervening.
None of this is solved by telling leaders to “execute harder”.
Performance improves when the organisation itself becomes better at making and carrying consequential decisions.
That means knowing where authority sits.
Giving accountability to people who can actually influence the outcome.
Creating information the organisation can trust.
Separating completed activity from realised value.
Protecting scarce leadership capacity.
Knowing when the centre should intervene.
And increasingly, determining which decisions can be delegated to technology without delegating responsibility with them.
The strongest enterprise is not the one in which senior leadership makes every important decision, nor is it the one in which everything has been decentralised.
It is the one in which consequential decisions are made at the right level, with the right information, by people or systems operating inside clear boundaries, and where somebody remains answerable for what happens next.
That is what turns leadership from individual capability into institutional capability, and that is what allows strategy to survive contact with the organisation expected to deliver it.
Before you change the structure, add another transformation programme or demand more accountability, establish how the enterprise is actually making and executing decisions.
E. Cartier & Co works with boards, chief executives, owners and senior leadership teams where strategic ambition is beginning to exceed the organisation's ability to execute it reliably.
The work begins with the decisions and outcomes that matter.
Who actually controls them? Where does information weaken? Which decisions are repeatedly escalated? Where is executive capacity being consumed? Which performance measures are genuinely comparable? Where does the formal organisation differ from how the company really operates? What should remain local? What belongs at enterprise level? What can technology decide? What should it never decide alone? And where has the organisation become dependent on an individual whose judgement has never been institutionalised?
The conclusion may involve organisation design.
It may not.
Authority may need to move closer to the customer. A group function may require greater control. Several transformation programmes may need to become one. Another may need to stop. A leadership role may be unnecessary. A new one may be critical. An AI deployment may be accelerated because the governance is stronger than management thought.
Another may need to wait, or the review may conclude that the structure is broadly right and the real problem lies somewhere else entirely.
There is no preferred organisational model before the performance problem has been understood.
That is the point.