We spent 3 hours arguing about hiring 4 or 6 people (and completely missed this)

Sponsored by Maxio

Want to learn how to create a financial forecast in 30 mins using AI?

Join me for a free masterclass on the 6th November where I will provide tactical advice on how you can save time (and make better decisions) using AI to create amazing financial forecasts.

Spaces are limited, so make sure to register on the link below now so you do not miss it.

Hello hello!

Let me tell you about the worst headcount planning meeting I ever sat through.

It was 3 hours long. We had 14 people in the room. And we spent the entire time arguing about whether the sales team needed 4 new Account Executives or 6.

Sales said they needed 6 based on pipeline growth. Finance said the budget only allowed for 4. HR said they could only recruit 3 anyway because of capacity.

It became a competition on who shouted the loudest.

But here's what nobody could answer: "If we hire 6 Account Executives, who else do we need to hire?"

Nobody knew. Because nobody had actually mapped out how one hire triggers three more.

They were planning based on emotion instead of using an accurate model...

We came away with more questions than answers.

Does this sound familiar?

If you've ever spent more time arguing about headcount numbers than understanding how your business actually scales, you're not alone. Most companies are still treating headcount planning like a shopping list.

They know what they want to buy. But they have no idea what it actually costs downstream.

But some companies figured out something smarter. And today, I'm going to show you exactly what they did.

The Educated Guess

Here's the thing about headcount planning that nobody wants to admit.

It's not really planning. It's an educated guess.

This is what happens.

Your VP of Sales comes back from a pipeline sales review and says, "We're 20% ahead of our new customer target. We need to add another 4 Account Executives in Q2."

You look at the budget and say, "We can afford maybe 2 more employees."

Sales pushes back. "You want us to look after 20% more customers with only 2 more people?! This is not realistic."

So you compromise. You agree to 3 Account Executives starting Q2. Everyone signs off. The plan goes into the board deck.

And then this happens:

March - Your top Account Executive quits
April - Customer Success starts struggling and churn increases
May - The product team delays the new product launch and you end up over-staffed

But here's what's worse.

Nobody saw any of this coming. Because your headcount plan was just a list of roles and dates.

3 Account Executives in Q2. That's what you planned for. But what you didn't plan for was:

How many Sales Engineers do those 3 Account Executives need for support?
How many Customer Success Managers do you need when those deals start closing in Q3?
When does the sales team get big enough that you need another Sales Manager?

Your plan doesn't answer these questions. Because it's not built to show how things connect.

And this is what makes headcount planning so impossible:

Everything is connected.

You can't hire Customer Success until sales closes the deals.
You can't hire junior analysts until the senior manager is in place.
You can't hire Sales Engineers if a product launch is delayed.

But most headcount plans are just static lists. Role, start date, salary, total cost.

There's no way to see what happens downstream. No way to answer: "If sales is ahead in Q2, what does that mean for operations in Q3?"

The real cost isn't just the time you spend updating your plan every week. It's the strategic decisions you're NOT making because nobody can see how hiring in one area affects three others.

You can't make good decisions from a static list that doesn't show how your business actually works.

And you can't show up to a board meeting with 12 educated guesses.

From Static Plan to Dynamic Model

So here's what the smartest finance teams figured out:

Stop treating headcount planning like a list.
Start treating it like a model.

What do I mean by this?

Traditional headcount planning is basically a static list. Role, start date, salary, total cost. You add up the numbers and hope they're right.

But a model is different. A model shows how things connect. A model lets you say: "If this happens, then that follows."

And this is exactly where AI becomes super powerful.

Let me show you what I mean.

Let's say your sales team is planning to hire 6 new Account Executives next year. In your old static plan, you'd just put:

6 Account Executives
Average salary: £80K
Start dates: Q1-Q4
Total cost: £480K

Done.

But this is not how it actually works.

Each Account Executive needs support. You probably need 1 Sales Engineer for every 5 Account Executives. And 1 Customer Success Manager for every 10 customers those Account Executives close. And if you're hiring that fast, you will need another Sales Manager by Q3 to manage the expanded team.

Your static list doesn't show any of this. But AI can help you build a model that does.

AI can help you:

Understand what connections exist between roles (even if you have no knowledge of how a department works)
Build a model you can understand, audit, and adjust
Map out how one hire triggers others downstream

You can tell the AI: "Each Account Executive closes an average of 8 deals per year. Each deal requires 0.1 Customer Success Manager capacity. We need 1 Sales Manager per 8 Account Executives. Show me the full downstream impact."

And suddenly you're not just planning for 6 Account Executives. You're planning for 6 Account Executives, plus 2 Sales Engineers, plus 1 Customer Success Manager, plus 1 Sales Manager.

That's 10 hires, not 6. And the total cost is now £720K, not £480K.

And this is super important: Most companies think their headcount problem is "We need better forecasts."

But this is not it.

The real problem is: "We need to understand how our business actually scales."

AI can help you build a model of how your business scales, so you can make better decisions about when and where to hire.

And once you have that model, everything gets easier.

You go from being reactive (updating your list every week) to being strategic (modelling different growth scenarios and making informed decisions).

How to Build Your Headcount Model (Without Spending Hours in Excel)

Here, I'm going to show you exactly how to solve the specific problem with headcount planning:

Using AI to understand how one hire triggers three more. Then building this into a model.

Here's the roadmap in three steps.

Step 1: Define Your Logic

Tool that will help: ChatGPT Branching

This is where you branch into different scenarios whilst maintaining your previous conversation. Think of this like keeping multiple versions of a Google Doc open at once - you can explore different approaches without losing your work, then pick the winner.

You can access it by clicking the three dots under your chat:

And then this will appear:

You can find out more here…

1. Upload your current headcount spreadsheet

Just drag and drop your Excel file into ChatGPT (make sure you are allowed to upload confidential data in to your AI tool, if not, use anonymised data).

2. Use this prompt:

You are a headcount dependency mapper. When I describe a hiring plan, you ALWAYS ask about downstream effects before building models.

Here's my current headcount data. I want to understand: if I hire [X] people in [Department], what other roles get triggered downstream?

Interview me, and ask me 3 detailed questions to map out the dependencies.

As a result, I want to create 3 scenarios, conservative, median and extreme

3. Let ChatGPT interview you

This is where it gets good. ChatGPT will start asking you questions like:

"How many Account Executives can one Sales Engineer support?"
"At what team size do you need an additional manager?"
"How long after an Account Executive starts do you typically need Customer Success Manager support?"

Answer honestly and briefly. If you don't know the exact ratio, give your best guess. You can refine this later.

Pro Tip - Ask ChatGPT to give you the top 5 questions that will allow it to produce the best possible results. You don’t want it to ask you 25 questions you do not know the answers to (or that takes you 4 hours to answer!).

4. Use branching to test different logic

Here's the next trick: after ChatGPT proposes something, don't just accept it. Click the "branch" button in the ChatGPT option menu.

How to branch into a new chat within ChatGPT

Create three branches:

Branch 1: Conservative ratios
Branch 2: Median ratios
Branch 3: Aggressive ratios

In each branch, ask:

Based on my [conservative/median/extreme] ratios, if we hire 6 Account Executives next year, what's the full downstream headcount plan?

Compare the three outputs side by side. Which one feels most realistic for your business? That's your logic.

5. Export your winning logic

Once you've picked the branch with the logic that you think most closely follows your business, ask:

Great, now I want you to turn your response into logic I can use when a creating a model. Please simply output:

1. All business drivers (what triggers hiring?)

2. All dependency ratios (1:X relationships)

3. Timing offsets (how long between triggers?)

4. Any constraints (budget, manager capacity, etc.)

Do not over-complicate this.

Copy this output. You'll need it for Step 2.

Step 2: Create the Model

Now we're going to turn that logic into something you can actually use.

Start a new ChatGPT conversation. You have two paths here. Pick based on your resources and how you work:

---

Path A) Use ChatGPT to Generate Excel with Formulas

Use this if you want a complete Excel model with all formulas built in 5-10 minutes. This is the approach I show in my video "How to Use ChatGPT 5 to Build INSANE Financial Models."

Here's what you do:

Open ChatGPT and activate thinking mode (top left). Really important.

Upload your current headcount data and paste in the logic you exported from Step 1.

Then use this prompt:

I'm a CFO and I need to prepare a headcount plan with three scenarios.

Scenario 1 (Base): Hire 6 Account Executives over the year. Apply standard dependency ratios.

Scenario 2 (Growth): Hire 8 Account Executives with accelerated timeline (faster ramp).

Scenario 3 (Conservative): Hire 4 Account Executives with extended timeline.

Using the logic below, build me a financial model in Excel with formulas. Really important: I want an assumptions sheet where I can change parameters which will influence the model.

Use my current headcount data as the baseline [Attach spreadsheet].

# Logic

[Enter logic from previous step]"

ChatGPT will think for a while. Then it will give you a downloadable Excel file.

What you'll get:

An assumptions sheet with all your ratios and parameters
One sheet per scenario showing the full headcount cascade
Everything linked with formulas
A readme explaining how the model works

Open the file. Test it. Change an assumption in the assumptions sheet. Watch the scenarios update automatically.

This is a working model with real Excel formulas. Not a secret black box that only AI knows how it was generated. You can see every calculation, audit it and change it.

The advantage: This takes 5-10 minutes. You get a complete, formula-driven Excel model that anyone on your team can use.

What to check: Review the model like you would review work from someone on your team. Check the formulas. Make sure the logic makes sense. Test the dependencies. AI did the work, but you're still responsible for validating it.

---

Path B) Use Excel Agent Mode (Assuming you have Copilot for M365)

Use this if you want Excel to build the model for you automatically with even more sophistication. This uses Excel's new agent mode.
You can learn more in my video here.

What is Excel Agent Mode?

It's an AI built into Excel that can build financial models in just a few minutes.

Here's how:

1. Get Excel Agent Mode

You need Excel Labs. In Excel online, go to Add-ins → More Add-ins → Search for "Excel Labs" → Install "Agent Mode Frontier."

This requires Microsoft 365 with Copilot license and you need to enable the beta channel.

If you have difficulties getting Excel Agent Mode, here you will get the most recent information from Microsoft on how to get it.

2. Describe what you want

Click the Excel Labs button. Paste in your headcount logic from Step 1. Then describe your model:

Build a headcount model over 12 months with dependencies. I want to hire 6 Account Executives across Q1-Q4. Model the downstream impact based on these ratios: [paste your logic].

Create separate tabs for: Base hiring plan, Downstream roles triggered, Monthly cost breakdown, Assumptions I can change.

Include formulas so when I change assumptions, everything updates.

3. Watch it build

The agent will think through this. You'll see it creating tabs, building formulas, adding documentation. It will even fix its own calculation errors as it goes.

4. Test and refine

Change your assumptions. Hire 8 Account Executives instead of 6. Watch the whole model recalculate. Add a new dependency by selecting a cell and asking the agent to adjust the logic.

The advantage: Even more automated than ChatGPT. Builds directly in Excel with sophisticated formulas and documentation. Can iterate and fix itself.

The limitation: Requires Microsoft 365 Copilot license and Excel Labs access. Not everyone has this yet.

Step 3: Run This Live in the Boardroom

This is where everything becomes real.

You walk into your board meeting. Someone says: "What if we grow 20% faster than planned in Q2?"

If you used Path A (ChatGPT-generated Excel):

Pull up your Excel file. Share your screen.

Go to the assumptions sheet. Change "Q2 hires" from 2 Account Executives to 3. Change "growth rate" to 20%.

The formulas recalculate instantly across all scenarios.

If you used Path B (Excel Agent Mode):

Same approach. Change the numbers in your Excel model. Everything recalculates.

Or ask the agent: "Model a scenario where we accelerate Q2 hiring by 20%." It will create a new scenario tab for you right there in the meeting.

Doing this, you go from being the person who reports what happened to being the person who shapes what happens next.

This is the difference between spreadsheet planning and model-based planning.

The Bottom Line

Headcount planning fails when you treat it like a spreadsheet exercise. It works when you treat it like a modeling problem.

Here's what that means:

Map your dependency logic - Use ChatGPT to understand how one hire triggers three more.
Build your model - Either use ChatGPT to generate Excel with formulas, or use Excel Agent Mode for more automation.
Test it live - Use it in real meetings to answer "what if" questions in 60 seconds, not next week

Why this changes everything.

Scenario planning becomes instant.

"What if sales is 20% ahead of plan in Q2? What does that mean for our Q3 hiring?" Instead of spending 2 hours rebuilding your spreadsheet, you get the answer in 30 seconds.

Dependencies become visible.

"We can't hire these 3 junior analysts until we hire the senior manager. And that manager can't start until Q2 because we're still defining the role." Your model shows this automatically.

Budget conversations get smarter.

Instead of arguing about whether you need 6 or 4 Account Executives, you're discussing: "If we hire 6, here's the total downstream impact. If we hire 4, here's how that changes the support team and the timeline."

This isn't about AI predicting the future. It's about building a tool that helps you see the downstream effects of your decisions faster, so you can actually make those decisions instead of just reporting on them.

Your Move

Resources to get you started:

Think about your next board meeting.

The one where someone will inevitably ask: "What if we increase hiring in Q2?" or "What happens if we do not hire, but still need to meet the growth target?"

You have two options.

Option one: You explain that you'll need to go back, update the assumptions, and have an answer by next week.

Option two: You pull up your model. You change the inputs right there in the room.

So here's my question: Which one of those finance pros do you want to be?

Take 30 minutes this week. Pick one department. Build a simple model using whichever path makes sense for you.

Test it in your next planning conversation. See if it changes how the conversation goes.

Imagine your work without delays. Just better decisions..

This is the power of AI.

Best,

Your AI Finance Expert,

Nicolas

P.S. - What did you think of this approach? Hit reply and let me know if you're planning to try this for your team (I read all replies).

P.P.S. - If you want to learn from me on how to use AI for forecasting, join my next masterclass sponsored by Maxio. It's free and I have prepared 3 practical use cases! Register here for free

Tips & Insights on Finance & AI

We spent 3 hours arguing about hiring 4 or 6 people (and completely missed this)

The Educated Guess

From Static Plan to Dynamic Model