Recently the CEO of my client – one of the most well-known retailers in the world – told my team that he wanted to see his organization data “de-averaged” as part of a benchmarking effort he had commissioned. I was ecstatic. The tyranny of averages plagues us all! How refreshing that our enlightened CEO understands how the average number of direct reports at other companies is not at all helpful in driving good design decisions for his organization. Now, we organization design folks know that any company’s structure is a collection of decisions made over time using a wide array of inputs. The strategic rationale based on unique department needs, legacy leaders reluctant to let go of their loyal followers and building structure around them, people with unique skills and expertise that can lead more than others, dramatic cost reduction requirements to keep the quarterly reports looking viable for the street, and so on. In other words, getting an average of span of control or an average headcount number tells you an average that was arrived at through a complex series of choices and circumstances that you know nothing about.
The Ambiguity of Benchmarks
How are span of control and sizing benchmarks calculated? The scores that make up the benchmarks are simply the average scores for the group you are comparing to. Somehow, comparing performance to a benchmark sounds as if we are setting a higher “bar” than comparing to any average, but we are not. No matter where you look and how you benchmark, the average span of control is between six and eight, however the range is between 1-100. As you may recall from your statistics classes, an average is simply a typical value in a set of data, and statistics with lower variances are better. All that to say, these are not even very good statistics. But this argument often falls on deaf ears. Why? Over the years I have come to understand that when a client asks for organization design benchmarks, they are really trying to figure out how many resources they should have, doing what kind of work and delivering what kind of outcomes for what kind of value. Clients look to benchmarks to help to answer these questions, presumably because they are unsure how to use the correct set of design inputs to determine the answer. Let’s say you ask for finance organization benchmarks. I can tell you the average number of people some other organizations have in finance and the associated average cost, as well as the average number of direct reports. But how does that really help you? You don’t know:
- What services that function provides
- How good the services are
- Whether or not their functions are largely embedded in the business (thus only a lightly resourced COE model is funded)
- What technology they have that automates their work
- If they are a single product org in one geography or a global, multi service organization, which requires more resources and more span breakers on time zones
- What portion of their workforce consists of contractors, consultants, outsource vendors and gig workers (which likely won’t show up in any of the numbers)
In other words, even with a traditional approach to benchmarking, you don’t know anything you need to know to use these averages to guide any of your organization design decisions.
Organization Design Is More Than a Headcount
Organization Design is more than headcount. It is roles and responsibilities, networks of teams, metrics and rewards, and decision rights and governance. But time and again, we are asked for spans and layers benchmarks. So here is a summary for all.
| Level of Managerial Position | Span (Number of Direct Reports) |
| First line supervision: | |
| Technical positions | 6-8 |
| General/sales positions | 4-8 |
| Clerical positions | 10-15 |
| Field positions | 12-25 |
| Middle/upper management | 5-9 |
| Senior/executive management | 4-8 |
Layers of Management The following guidelines are suggested for layers including the level of officer for each department. Global Centralized Functions should consist of no more than five layers. Geographically Dispersed Businesses should include no more than six layers
Based on the insights above, are your organization design decision made? I think not. Perhaps you too will agree that benchmarks are mildly interesting at best, and highly misleading at worst. If your competitor has a slightly higher or lower span of control than you, what does that tell you to do to improve your design to achieve better business results?
Asking the Right Questions
Instead of the tyranny of averages, you must consider many factors in concert to arrive at an organizational model and structure right sized for your business. The questions to start with include:
- What is your company’s purpose, strategic intent, and place in the ecosystem?
- What is the optimal business model to support your purpose?
- Where should you standardize as consistency is more valuable?
- Where should you encourage innovation as agility is more valuable?
- What capabilities should be cultivated within the organization, what should be outsourced or contracted?
- Where do we need integration, and where do we want separation?
- Where do we want our teams to feel their primary sense of belonging, their “home base”?
At EY, we have taken a new approach to “benchmarking.” We use our Organization and Workforce Analytics Platform to provide an outside-in view. We look at the strategies our clients use to win in their respective industries. We describe the choices they made and why they made them based on their specific set of circumstances. Then we follow the same approach to inside-out analysis for our unique client situation to arrive at the right structure, integrated networks of teams, KPI’s and incentives, and governance and accountabilities. There are many compelling insights you can glean from examining your organization data beyond benchmarks. In my next article, I will describe more detail on spans and layers and why the right design has a strong positive correlation to organization performance and employee experience.
Order Strategies for Organization Design Below!
