#76 JooBee's newsletter

TL;DR

🕰️ Org design over 100 years. What changes with AI?

🏗️ How to design an organisation in 4 steps 

🎧 Listen to the newsletter here

This newsletter edition is brought to you by Rippling 💛

100 years of org design. What changes with AI? 

I get asked a lot: “How do you do organisation design?”

Lately, the question has grown a new ending: “…especially now with AI?”

And my answer is perhaps less exciting than people expect… 😅

A hundred years of how we designed organisations 

I come from a Systems Engineering background, so I can't help seeing an organisation as a system: something built to achieve an outcome, shaped by different variables (parts) that affect how you design it.

AI is a significant variable, perhaps the most significant of this era. But it's one in a looooong line. A few highlights from the last 100 years…

Whenever a new variable changed what was possible, organisations changed their design in response.

The first principle of org design: balance value and sustainability

One key thing I want you to notice, if you strip away what has changed, every organisation design (or redesign) is trying to do the same thing: 

I think of these as the two boundaries wrapping around the system (a.k.a. the organisation). 

• On one side: Value. What does the organisation need to do to create value for its customers?

• On the other: Sustainability. How do we keep doing that without the cost becoming unsustainable? 

Everything in between is what we get to design: the people, teams, layers, technology, processes, capabilities, data and decision-making.

When you design a system, move one variable and it affects another (that's why they're called variables 😅). Add a layer and you might get better decisions, but you've also added cost. Split a team for greater focus and you've created another handoff. Centralise for consistency and you've moved decisions further away from the work.

AI gives us a significant new variable to work with. It changes what work people can do, what technology can do, the capacity of a team and the capabilities we need, and even how many people, managers or layers we need. 

But the system still needs to balance creating value with the cost of delivering it. 

Keep the purpose constant as you evolve the design 

We can debate spans, layers, reporting lines, decision rights, management ratios and now where AI can replace or augment work. 

But we're already too far downstream if we haven't first asked: How does the organisation create value for its customers? 

Designing an organisation without understanding how the business creates value is like designing a restaurant kitchen without knowing the menu. You could build the perfect bakery, only to discover you're serving sushi 😱.

AI changes the equipment. It doesn't decide what you're cooking. The business is still the starting point.

How to design an organisation in 4 steps

A recent founder client engaged me just after raising their Series A. They were getting ready to scale and were adamant about one thing: “In month one, we need to sort out the org structure.”

Then we had our first kick-off, and I started with the business. 

They quickly realised that their most important next steps had nothing to do with people or structure. They needed clarity on the business strategy first — what they were building towards, the bets they were making and what needed to be true to create value for their customers.

As we explored different bets and scenarios, I visualised what the organisation could look like for each (not as solutions, but as possibilities). It became clear that, until they made those business decisions for the next 12–24 months, we couldn't know what organisation they needed to deliver them.

That's why I never start organisation design with the org chart. I start with the business.

Here are the 4 steps I use to design an organisation 

1️⃣ Start with what the business needs to achieve

I take time to understand the business strategy. What value are we creating? For whom? How does the business make revenue? What are we trying to achieve over the next 12–24 months? What needs to be true for us to win?

Yet I've seen plenty of organisation design exercises start with: “We currently have 73 people. How should we organise them?” Or organisation design exercises that start with leaders listing the roles they want to hire next year 😱😱😱.

That's already too far downstream. Those people and proposed hires are solutions. First, understand what the business is trying to achieve.

2️⃣ Identify what work needs to happen

I want you to forget jobs for a moment. Forget titles. Forget teams. Forget who currently owns what. This step is simply about identifying the work that needs to happen for the strategy to succeed. 

What needs to be built, sold and delivered? What decisions need to happen? What information needs to move? Where are the dependencies and bottlenecks? What work are we doing today that we no longer need?

This is workload forecasting. For now, capture the work that needs to be done (and what doesn't!). That's it. Don't solve for who or how yet.

3️⃣ Find the best way to get that work done 

A common mistake is to identify the work and then jump straight to “Who do we need to hire?”, skipping the critical question: “What’s the best way to get this work done?” 

I've been banging this drum for a while. Back in 2024, I wrote that before forecasting roles and headcount, we should consider other options. Can we stop doing it? Simplify it? Redesign the process? Automate it? Use AI (yes, AI gets a tiny mention because it’s not the star of the show 😏)? Use AI to increase someone's capacity? Give it to an external partner? Develop someone internally? Or do we genuinely need to hire?

And don't assume the goal is automatically fewer people.

• Perhaps ten people become six.

• Perhaps ten people stay ten but can now serve twice as many customers.

• Perhaps fixing a terrible process creates more capacity than introducing any new technology.

• Perhaps technology removes execution work but leaves work requiring more judgement or customer interaction.

• Perhaps an entire workflow disappears.

The goal is simply to find the best way to get the work done. 

4️⃣ Then design the organisation around the work

Only now do I start answering the questions we normally associate with organisation design.

• Capabilities: What knowledge, skills and judgement do we need? Where do those capabilities need to sit? 

• Process and coordination: What needs to sit together? What should be centralised? Where should decisions happen? How should information flow? Where do we genuinely need management?

• Structure: What roles do we need? Who owns what? Who reports to whom? How many layers? How many people?

At this point, spans and layers become useful, and the org chart becomes meaningful because it’s designed around the work. 

Start from first principles and you can design any organisation 

These 4 steps are why I think about organisation design as system design, starting from first principles. 

The variables will keep changing, economics, society, technology, AI and whatever comes next. Each might change the organisation you end up with, but the way you work it out remains useful. 

Go back to the same 4 questions: 

1. What is the business trying to achieve?

2. What work does that require?

3. What’s the best way to get that work done?

4. What organisation does that work require?

That’s how you design the organisation the business actually needs.

What do you need to STEP UP next?