Key takeaways
AI increases managerial capacity by reducing administrative work, but it does not reduce the need for leadership, judgment, or accountability.
Leadership load, not administrative capacity alone, should determine the appropriate span of control.
Organizations should treat wider spans as an outcome of effective organizational design, not the objective
Artificial intelligence is prompting organizations to revisit one of the oldest questions in organizational design: How many direct reports should a manager have? As AI automates reporting, coordination, and information management, many organizations assume the logical outcome is wider spans of control and fewer management layers. It is a reasonable conclusion – but it is also incomplete.
In practice, spans of control have often been shaped first and foremost by managerial bandwidth: how many people a manager can coordinate, monitor, and support within the time available. AI expands that bandwidth by taking on much of the reporting, information gathering, and routine coordination that has historically consumed it. As this constraint recedes, the other determinants of span – leadership demands, work complexity, and risk – become more important, not less.
AI doesn’t make spans of control obsolete. It changes what should determine them.
Administrative capacity has never been the whole story
For much of the modern organization’s history, a manager’s available time has been the most visible and practical limit on span. Managers gathered information, prepared reports, coordinated activities, approved routine decisions, and ensured that work moved across organizational boundaries. The more time those activities required, the fewer people a manager could effectively oversee.
AI materially changes that constraint. It can synthesize information, identify exceptions, prepare updates, coordinate workflows, and make performance more transparent. In many roles, this will create real capacity for managers to support larger teams. But administrative capacity has never been the whole story. A manager may have more time and still face substantial demands for coaching, judgment, collaboration, and accountability.
A better way to think about spans
We find it helpful to think about a leader’s overall Leadership Load: the total demands placed on a manager by the people, work, decisions, and administration within the role. Leadership Load is driven by four primary factors:
– Leadership Requirements – the coaching, communication, performance management, and development the team requires.
– Work Complexity – the ambiguity, interdependence, and variety involved in getting the work done.
– Decision & Risk Load – the volume and consequence of decisions, including the commercial, operational, and regulatory risk carried by the role.
– Administrative Load – the reporting, coordination, scheduling, approvals, and information management required to keep work moving.
AI will not change all four factors equally. Its most immediate impact is on Administrative Load. It may also reduce some elements of Work Complexity by making information easier to access and routine decisions easier to support. But it does much less to eliminate the need for coaching, judgment, relationship management, or accountability. Those demands remain – and in some roles may become more prominent as administrative work recedes.
The leadership load framework
Wider spans are a consequence – not the objective
None of this suggests that spans of control will remain unchanged. Quite the opposite. The effect, however, will vary considerably by the nature of the role.
Consider a customer service operation in which agents follow well-defined processes, handle similar types of interactions, and escalate a relatively small number of exceptions. Today, a manager may spend a meaningful portion of the day compiling performance data, reviewing queues, distributing work, preparing coaching notes, and following up on routine issues. AI can automate much of that activity while also surfacing the employees and interactions that genuinely require attention. The manager is not simply working faster; the supervisory model itself changes. In this environment, a substantially wider span may be both practical and desirable.
Now compare that with a product development team. AI may help prepare analyses, summarize customer feedback, generate early concepts, or accelerate elements of the development process. But the manager still has to resolve competing priorities, integrate different technical and commercial perspectives, coach employees through ambiguous problems, and make trade-offs for which there may be no objectively correct answer. The team’s output may increase considerably without a corresponding reduction in Leadership Load. A wider span may be possible, but the increase is likely to be much more modest.
The distinction is even clearer in senior sales leadership. AI can improve forecasting, summarize account activity, identify pipeline risks, and reduce the administrative burden of managing a sales organization. In theory, that could allow one executive to oversee more sales leaders or even own every major customer relationship. But doing so may concentrate significant commercial risk, customer relationship risk, decision-making authority, and institutional knowledge in a single role. The question is no longer simply whether one leader can supervise more people. It is whether the resulting level of risk remains appropriate for the organization.
These examples point to the central design implication. Where work is standardized, teams are relatively self-sufficient, and exceptions can be identified reliably, AI is likely to support wider spans meaningfully. Where performance depends on apprenticeship, cross-functional judgment, complex trade-offs, or the distribution of material risk, the effect on spans will be smaller – even when AI produces substantial gains in individual productivity.
AI Will Change Spans of Control – But Not Equally
Conclusion
AI will change the economics of management. It will remove administrative work, make performance more visible, and give many managers the capacity to support larger teams. Organizations should take advantage of that opportunity. In functions where work is standardized and leadership demands are relatively low, wider spans and fewer layers may improve speed, accountability, and cost.
But applying that logic uniformly would repeat an old organizational design mistake: treating span of control as a benchmark rather than an outcome of the work. The right span will still depend on what employees need from their leader, how complex and interdependent the work is, and how much decision-making and risk the role carries. As AI reduces Administrative Load, these other factors should play a larger role in determining the answer.
The organizations that benefit most will therefore ask a more disciplined question than, ‘How many more people can each manager supervise?’ They will ask, ‘What Leadership Load does this role carry, and what span of control does that load support?’ AI doesn’t eliminate the need to answer that question. It makes answering it well more important.
Contact us for assistance designing your organization to maximize spans of control.
Frequently asked questions
1. Will AI eliminate the need for managers?
No. While AI reduces administrative work, managers remain essential for coaching, decision-making, collaboration, and accountability.
2. Should every organization increase spans of control because of AI?
No. The appropriate span depends on leadership requirements, work complexity, decision-making, and organizational risk, not AI alone.
3. How should organizations determine the right span of control?
Evaluate the leadership load of each role, including coaching needs, work complexity, decision and risk responsibilities, and administrative demands, before redesigning management structures.
Graeme Hartlen
Find out how we can use organizational design to create value in your organization.
CONTACT US
