Bahasa Melayu

Org Design: When to Split a Function, When to Combine

Key Facts: Org Design at Mid-Market Scale

  • Redesign cadence: McKinsey finds the average large company restructures every 2-3 years, but only ~23% of redesigns deliver their intended outcomes on first attempt.
  • Productivity cost: Gartner reports restructuring temporarily drops team productivity by 20-25% for 6-9 months post-announcement, regardless of whether the redesign is ultimately successful.
  • Span-of-control norms: Deloitte benchmarks show functional leader spans of 6-10 direct reports for strategic work, 10-15 for operational work, with quality erosion above 15 in knowledge-work functions.
  • Function splits vs. merges: Bain research finds 60% of structural interventions at 100-300 person companies are splits (growth-driven); 40% are combines (coordination-driven).
  • Reorg fatigue threshold: HBR data suggests employee trust in leadership drops sharply after a third reorganization within five years, independent of whether each reorg was individually justified.

The Coupling vs. Coordination Test

When two teams' work is tightly coupled — shared customer moment, shared outcome, shared decision surface — combine them under one leader, because separation creates handoff cost that no process can erase. When two teams' work is loosely coupled but requires expensive coordination overhead (meetings, escalations, conflict resolution) specifically because they share one leader whose attention is divided, split them, because independent leadership at the sub-function level produces better output than shared leadership at the combined level. The test is not size or seniority — it is whether the friction lives inside the work (couple it) or inside the coordination (split it).

At some point between 100 and 300 employees, almost every mid-market company faces the same structural question. The function that used to work (led by a single capable leader who could hold the whole thing in their head) isn't working the same way anymore. It's either grown too large for one leader to maintain quality, or it's become so fragmented that the handoffs between pieces are destroying the customer experience.

The question that surfaces is: should we split this into two separate functions with their own leaders, or should we combine these two related functions under a single unified leader?

And typically, the CEO and COO have the conversation for three months, go back and forth, have it again, and eventually make a call that feels arbitrary because there was no clear framework for deciding.

The reason it drags is almost always the same: the org problem and the people problem are entangled. Deciding whether to split Sales and Customer Success isn't just a structural decision. It's implicitly a decision about whether the current leader can run the combined function, or whether splitting creates two roles that fit two existing people, or whether combining requires a search for someone who doesn't exist yet. The three restructuring signals framework gives you the diagnostic to separate structural rot from a people problem before this conversation begins. Those questions are uncomfortable, and most CEOs delay the structural decision to avoid having the people conversation.

This framework untangles them.

Why This Decision Gets Delayed

Most split-vs-combine decisions involve someone's job. If you split the function, a single leader may become a leader of half. If you combine, one leader may lose their independence. These decisions have real human consequences, and decent CEOs feel that weight.

But delaying the structural decision to protect someone's current arrangement isn't kindness. It's the CEO absorbing a structural dysfunction to avoid a hard conversation. And the longer the structural dysfunction runs, the more damage it does.

There's also a diagnostic confusion. The CEO notices symptoms (cross-functional conflict, missed metrics, slow decisions) and can't tell whether the cause is structural or personal. If it's structural, changing the person won't fix it. If it's personal, changing the structure will just move the problem. Misdiagnosing which one you have leads to either an unnecessary restructuring that disrupts the organization, or a personnel change that doesn't fix the underlying structural flaw. McKinsey's research on organizational design identifies the personal/structural confusion as the leading cause of failed reorganization attempts at mid-market companies.

The framework below gives you criteria for the structural decision first. Then the people decision follows from it, not the other way around.

When to Split a Function

A function should be split when two conditions are met:

Condition 1: The function contains two sub-functions with fundamentally different success metrics, customer journeys, or time horizons that create irresolvable tension under one leader.

Irresolvable tension is the key test. Some functions contain natural tension that a good leader manages productively. Sales and Customer Success have different incentives around deal size and customer health, but that tension is manageable and sometimes valuable under one CRO who sets the right culture.

But when the tension is structural (meaning no amount of leadership quality can resolve it because the metrics themselves point in different directions), splitting is the right call.

Classic splits that work: demand generation and brand marketing (different time horizons: brand is 12-24 month payoff, demand gen is this quarter), enterprise sales and commercial/mid-market sales (different skills, different sales cycles, different management needs), product engineering and platform engineering (different customer focus: external product vs. internal infrastructure).

Condition 2: The function is large enough that one leader can't maintain quality across all sub-functions.

This is a span-of-control test. A leader with 8 direct reports in one function and 15 in another is probably managing the smaller one well and the larger one inadequately. A split lets each get the leadership attention it needs.

The threshold is roughly: when a functional leader has more than 15 people in their direct span of influence and the work requires detailed operational knowledge (not just strategic direction), span-of-control limits are being hit.

But span of control is not sufficient reason to split on its own. Splitting creates coordination overhead. Every split requires a new handoff protocol, a new shared-resource negotiation, and a new leadership relationship to manage. Deloitte's research on organization spans and layers finds that premature splits (driven by size rather than functional differentiation) tend to increase management cost without improving output quality. If the only reason to split is size, consider first whether the leader could hire a strong VP directly under them to handle the operational work.

The right test for splitting: would the two halves of this function actually perform better separately, or are you just trying to manage a size problem with structure?

When to Combine Functions

A function should be combined when two conditions are met:

Condition 1: The functions share enough customer context that separation creates duplication or handoff failures.

This is the opposite test from the split condition. When two separate teams are both touching the same customer moment (both have a view into the account, both make decisions that affect the customer experience) and they're coordinating poorly, separation is creating more friction than structure.

Classic combines that work: Customer Success and Professional Services under one VP when the handoff between onboarding (PS) and ongoing success (CS) keeps breaking. Sales and Partnerships under one revenue leader when partners are in every significant deal anyway. Product Management and Design under one CPO when the two are constantly blocked by each other's priorities.

The test: are the two functions spending more than 20% of their coordination time managing handoffs and resolving conflicts that exist specifically because they're separate? If yes, combining reduces overhead and creates cleaner accountability.

Condition 2: The company needs unified accountability for a single customer outcome.

This is the accountability test. When a customer outcome (revenue, retention, satisfaction, time-to-value) is split across two leaders who each own part of it, accountability is diffuse. When something goes wrong, there's a boundary dispute about whose responsibility it was.

Combining under one leader gives that leader clear ownership of the full outcome. They can make trade-offs between the sub-functions without escalating to the CEO. They can see the full picture of what's happening across the customer journey.

The test: if you ask "who owns [the customer outcome]," does the honest answer point to one person, or does it point to a partnership between two leaders? A partnership that works is fine. A partnership that routinely escalates to the CEO for resolution is a signal to combine.

The Four-Criteria Decision Matrix

Score each criterion 1 (no) to 3 (yes):

For a split: | Criterion | Score | |---|---| | Irresolvable metric/time-horizon tension between sub-functions | 1-3 | | Sub-function is large enough that span of control is creating quality gaps | 1-3 | | Strong candidates exist for both resulting leadership roles | 1-3 | | Handoff protocol between the two resulting teams can be defined clearly | 1-3 |

Total split score: Sum of four criteria (max 12). Score 10+ supports splitting. Score below 7 suggests keeping combined.

For a combination: | Criterion | Score | |---|---| | Customer handoff failures or coordination overhead is high between functions | 1-3 | | A single customer outcome requires unified accountability | 1-3 | | A leader exists who has depth across both sub-functions | 1-3 | | Combining reduces duplication without creating an unmanageable span | 1-3 |

Total combination score: Sum of four criteria (max 12). Score 10+ supports combining. Score below 7 suggests keeping separate.

The criterion "a strong candidate exists" (for split) or "a leader with depth in both exists" (for combine) is often the deciding factor in practice. The right structural answer becomes the wrong practical decision if there's no one to execute it.

Two Case Illustrations

Marketing Split at a 220-Person SaaS Company

A 220-person B2B SaaS company had a unified Marketing function under one VP who had been excellent when the company was at 80 people. At 220 people, Marketing was now 24 people, with two very different tracks: demand generation (paid, SDR pipeline, intent data, ABM) and brand/content/communications (brand positioning, content marketing, PR).

Demand gen was being measured weekly on pipeline contribution and MQL quality. Brand was being measured quarterly on category awareness and content engagement. The VP was spending 70% of her time on demand gen (because it was urgent and board-visible) and brand was drifting.

Running the split matrix:

  • Irresolvable tension: 3. The time horizons are fundamentally different. No leader can genuinely optimize for this quarter's pipeline and the two-year brand simultaneously.
  • Span of control: 3. 24 people with two genuinely different practices. The span is creating quality gaps in brand.
  • Candidates for both roles: 2. The existing VP was clearly the demand gen leader. Brand leadership required a search.
  • Handoff protocol: 3. Brand feeds SDR sequencing and content; the handoff is well-defined.

Total: 11/12. Clear split signal.

The company split the function. The existing VP became VP of Demand Generation. They ran a 60-day search for a VP of Brand and Content. Both functions improved within 6 months. The CEO's main reflection: "We should have done this 18 months earlier. The VP was stretched thin and didn't tell us because she didn't want to admit she couldn't manage everything."

Combining CS and Professional Services at a 150-Person Company

A 150-person B2B platform company had separate Customer Success and Professional Services teams. CS owned post-sale health and expansion. PS owned implementation and onboarding. The two teams were supposed to hand off customers after implementation was complete.

In practice, the handoff was the most common source of customer complaints. PS considered implementation complete when the technical configuration was done. CS received a customer who didn't yet understand how to use the product, hadn't changed their workflows, and was already a churn risk. The PS leader said it was a CS problem. The CS leader said implementation had been handed off too early.

Running the combination matrix:

  • Handoff overhead: 3. The handoff between PS and CS was the company's top customer experience problem.
  • Single outcome needing unified accountability: 3. Customer time-to-value needed one owner.
  • Leader with depth in both: 2. Neither current leader had depth in both areas. A VP-level search was required.
  • Span manageability: 3. The two teams together were 18 people, manageable for a strong VP.

Total: 11/12. Strong combination signal.

The company combined CS and Professional Services under a newly hired VP of Customer Experience who had run similar combined functions before. Hiring this type of cross-functional leader follows a similar process to hiring a CS leader, but the criteria must include demonstrated experience owning the full customer journey, not just success post-onboarding. The combined team redesigned the customer journey as a single experience without a formal handoff. Customer time-to-value improved from an average of 90 days to 45 days within 8 months.

The Org Design Brief Template

Once the decision is made, communicate it through a structured brief to the leadership team:

What's changing: [Describe the structural change in one paragraph]

Why now: [The specific signals that made this the right moment: use the diagnostic criteria, not just "we decided it was time"]

Who leads the resulting function(s): [Named leaders, or explicit timeline for search if search is required]

What the new accountability structure looks like: [Who owns what outcome. One sentence per leader.]

What stays the same: [What the affected teams can rely on not changing: processes, customers, team membership unless noted otherwise]

What changes for each team: [Specific, honest list of what will be different for each affected team]

Timeline: [When does the change take effect? What are the transition milestones?]

The "what stays the same" section is critically underused. When a restructuring is announced, people assume everything is changing. Explicitly naming what isn't changing is as important as naming what is, and it reduces the uncertainty that causes unnecessary attrition.

The People Entanglement

The structural decision needs to come first. Once you know whether to split or combine, the people question becomes: who is the right leader for the resulting structure? Harvard Business Review's research on organizational structure and accountability makes the same point: the decision rights architecture must be designed independently of the personalities who will inhabit it, or the structure ends up perpetuating the people problem it was meant to solve.

That question may have a clean answer (the current leader is right for one of the resulting roles), an uncomfortable answer (neither current leader is right for the new structure), or an external answer (the right leader doesn't exist internally and requires a search).

All three are manageable. The worst outcome is reversing the logic: keeping the current people arrangement and designing the structure around it. That produces structures that fit the people rather than the work, and it defers the people decision into an increasingly uncomfortable future. AI is adding a new variable to this calculation: as AI tools remove coordination work from individual roles, the span of control thresholds that justify a split are shifting. Some functions that needed splitting at 15 reports can now be managed effectively at 20+.

How Rework Helps Track Cross-Functional Work After a Restructure

The hardest part of a split-or-combine decision is not making the call — it is the 6-9 months afterward, when new handoffs are still being defined, accountability boundaries are still being tested, and the productivity dip Gartner flags is actively eroding confidence in the decision. Rework Work Ops (from $6/user/month) is built for exactly this window. When you split a function, Work Ops gives each new team its own workspace while preserving shared views across the former unified function — so the new VP of Brand and the new VP of Demand Gen can operate independently without losing visibility into what the other is shipping. When you combine functions, Work Ops consolidates previously-separate project trackers, approval flows, and OKRs into a single source of truth under the new unified leader, reducing the "where do I find this" friction that follows every merger. Cross-functional work items carry explicit ownership tags, so the old PS-vs-CS boundary disputes get replaced with named accountability. For CEOs running the 18-person combined function or the two 12-person split functions, this is the operational layer that makes the structural decision legible.

Frequently Asked Questions

Learn More