Financial Scenario Planning: How FP&A Can Build Scenarios That Actually Help Management

I have trust issues with scenario models.

Not because I don’t like scenarios.

I do.

I’ve built plenty of them.

I’ve also sat in meetings where we had so many scenarios that I started to suspect we were using Excel to avoid making a decision.

You know the meeting I’m talking about.

Base case.

Upside case.

Downside case.

Management case.

Board case.

Sales case.

Recession case.

Soft landing.

Hard landing.

Something called “Adjusted Base” that apparently became necessary after Tuesday.

At some point, somebody asks which scenario we actually believe.

And the room gets very interested in the spreadsheet.

That’s usually where I start paying attention.

Scenario Planning Is Having a Moment

There’s a reason FP&A teams are talking more about scenario planning. A scenario is not a replacement for the forecast; it’s a way to test the assumptions around it. If the underlying forecast process is weak, start with the FP&A forecasting process first.

Businesses are dealing with enough uncertainty that a single forecast can feel almost silly. Current FP&A guidance is increasingly pushing companies toward continuous planning, predictive analytics, and faster scenario modeling rather than relying on one static view of the future. IBM

AI is accelerating that too.

Generating another scenario used to require an analyst to change assumptions, check formulas, update outputs, make sure nothing broke, and eventually wonder why they chose Finance.

Now technology can generate scenarios much faster.

That’s useful.

It also removes one of the natural constraints we used to have:

Scenarios were annoying to build.

I’m serious.

There was some value in that.

If you knew building another scenario would consume half your afternoon, you generally wanted a reason for it.

Now we can produce 40.

I’m not convinced we need 40.

Let’s Look at Your Forecast for a Second

Say your current revenue forecast is $100 million.

Management wants to understand the range of possible outcomes.

Perfect.

You create:

Base: $100 million

Upside: $110 million

Downside: $90 million

Nothing wrong with that.

But I’d ask you something before we go any further:

Why is the upside $110 million?

Don’t open Excel yet.

Tell me.

What has to happen in the business for that number to become true?

Maybe pipeline conversion improves.

Maybe pricing holds.

Maybe customer churn falls.

Maybe a product launch performs better than expected.

Maybe Sales hires six people earlier than planned.

Good.

Now we’re getting somewhere.

Because $110 million isn’t really the scenario.

Those things are.

A Scenario Isn’t a Number

This is where I think Finance occasionally gets itself into trouble.

We think:

Base / Upside / Downside

are scenarios.

They’re outputs.

A real scenario is a coherent set of things happening in the business.

Imagine a SaaS company.

Your upside scenario assumes:

higher pipeline conversion,

lower churn,

faster sales hiring,

and stronger pricing.

Okay.

Why are all four improving at the same time?

Maybe there’s a reason.

But I’d like to hear it.

Because sometimes an upside scenario quietly becomes:

Everything good happens simultaneously.

And the downside scenario becomes:

Everything bad happens simultaneously.

I have enough natural pessimism to appreciate the second one, but it’s not necessarily analysis.

The interesting scenarios are usually messier.

Maybe demand stays strong but hiring slips.

Maybe revenue grows while margins compress.

Maybe volume falls but pricing holds.

Maybe Sales hits the number but collections slow down.

Now we have something worth discussing.

Businesses rarely have the courtesy to make every variable move in the same direction.

Ask What Changed in the World

Here’s how I’d want you to build the scenario.

Start with the event.

Not the spreadsheet.

Suppose you’re worried about a recession.

Fine.

What actually changes for your company?

Customers delay purchases?

Sales cycles get longer?

Discounting increases?

Churn rises?

Hiring slows?

Vendors become more negotiable?

Interest expense stays elevated?

Some of those might happen.

Some might not.

Now translate those changes into drivers.

Then translate the drivers into financial outcomes.

Event → Business behavior → Driver → Financial impact

That’s the chain I care about.

If you skip the middle and simply tell me:

“The recession case is revenue down 15%.”

I’m going to become annoying.

Why 15%?

What changed?

Which customers?

Which products?

When?

Does margin move too?

What happens to cash?

Does management respond?

Because management is allowed to respond.

That part gets forgotten surprisingly often.

Your Downside Scenario Should Include You

This is one of my bigger complaints with scenario models.

They often model what happens to the company.

They don’t model what the company does about it.

Let’s say revenue starts falling.

Does the business keep hiring exactly as planned?

Probably not.

Does Marketing keep spending exactly as planned?

Maybe.

Does the company delay capital spending?

Reduce contractors?

Change pricing?

Renegotiate vendors?

Push a product launch?

Draw on its revolver?

I don’t know.

That’s why we’re doing the exercise.

A downside scenario where management sits perfectly still while the building burns is certainly one possible scenario.

It just shouldn’t automatically be the only one.

I’d want to see:

What happens if this occurs?

Then:

What would we do?

Then:

What does the business look like after we do it?

That’s much closer to decision support.

More Scenarios Can Actually Make the Meeting Worse

This is the part where scenario planning becomes a little dangerous.

Let’s say Finance brings five scenarios into the meeting.

Management can probably discuss five.

Now let’s say AI generates 75.

Technically impressive.

What exactly would you like the CFO to do with them?

Current FP&A thinking is moving toward AI-assisted scenario generation and continuous planning because technology can evaluate more possible futures much faster.

That’s potentially valuable.

But computational capacity and management attention are not the same resource.

Your model may be capable of evaluating 10,000 combinations.

The executive team still has Tuesday from 2:00 to 3:00.

Probably 2:07 to 2:52 once everyone arrives and somebody asks whether we’re waiting for Mark.

Finance’s job isn’t to show management every future the machine can imagine.

It’s to identify which futures matter.

I Want Scenarios That Produce Different Decisions

This is my favorite test.

Put your scenarios next to each other.

Then ask:

Would management do anything differently in each one?

If the answer is no, I’m not sure why I need all of them.

Suppose you have:

Scenario A: Revenue $100M

Scenario B: Revenue $98M

Scenario C: Revenue $96M

And management’s response in all three cases is:

Keep operating exactly as planned.

You may have three forecast outputs.

You probably don’t have three useful management scenarios.

Now suppose:

At $96M, hiring pauses.

At $92M, discretionary spending gets reduced.

At $88M, capital projects are delayed.

At $84M, liquidity becomes the priority.

Now we’re talking.

The numbers are connected to actions.

That’s what makes scenario planning useful.

Find the Thresholds

If I were sitting next to you reviewing your model, this is where I’d spend time.

Not on whether the downside is precisely $88.3 million.

I’d ask:

Where does management behavior change?

Where do we stop hiring?

Where does cash become uncomfortable?

Where do we violate a covenant?

Where does gross margin create a problem?

Where does headcount become too heavy for the revenue base?

Where would we change pricing?

Where would we reconsider an investment?

Those are thresholds.

And thresholds are much more interesting than arbitrary percentages.

Your business probably doesn’t care that revenue declined exactly 10%.

It cares that when revenue reaches a certain level, something else becomes true.

That’s the number I want Finance to find.

Your Scenario Model Needs Triggers

Once you find the threshold, give management something observable.

Let’s say your downside scenario assumes revenue will weaken because pipeline conversion deteriorates.

Great.

What are we watching?

Maybe:

conversion below 22% for two consecutive months.

Maybe it’s bookings.

Maybe churn.

Maybe backlog.

Maybe customer traffic.

Maybe utilization.

Maybe something completely specific to your business.

The point is that you shouldn’t have to wait until the P&L confirms the downside scenario six weeks after it has already started happening.

By then Finance is doing archaeology.

Useful, occasionally.

Not what I want from a forecast.

I want to know what signal tells us:

We are moving from this scenario into that one.

Then management already knows what conversation comes next.

The Base Case Is Usually the Most Political Scenario

Nobody says this out loud enough.

The downside case can be pessimistic.

That’s its job.

The upside case can be optimistic.

Also its job.

But the base case?

Now everyone has feelings.

Sales has a number.

The board has a number.

The CEO has a number.

Finance has a number.

The budget may have a number left over from six months ago wandering around the company looking for relevance.

And somehow we’re supposed to call one of them the “most likely.”

This is where FP&A earns its keep.

Don’t ask:

Which number will make everyone comfortable?

Ask:

Which set of assumptions do we actually believe today?

That’s different.

Sometimes the answer won’t match the target.

That’s okay.

A forecast isn’t supposed to flatter the target.

It has other work to do.

Don’t Let Probability Make Bad Assumptions Look Scientific

I’m fine with probabilities.

Used thoughtfully, they can help.

But I’ve also seen percentages give assumptions an undeserved air of sophistication.

Base case: 60%

Upside: 20%

Downside: 20%

Okay.

Where did 60% come from?

If there’s actual statistical work behind it, wonderful.

If someone stared at the scenarios for a while and thought 60 felt appropriately responsible, let’s call it what it is.

Judgment.

Judgment is allowed.

Finance uses judgment constantly.

Just don’t put a percentage sign on an opinion and expect me not to notice.

Excel has many powers.

Turning confidence into evidence isn’t one of them.

What I’d Actually Put in Front of the CFO

I don’t want a 47-tab scenario model in the meeting.

Build it if you need it.

Enjoy yourself.

But what I want management to see is much simpler.

For each meaningful scenario:

What happened?

Which assumptions changed?

What’s the financial impact?

What signals tell us it’s happening?

What would management do?

When would we act?

That’s enough for a very good conversation.

Notice what isn’t on that list:

Every permutation Finance managed to calculate.

The model supports the decision.

It isn’t the decision.

AI Is Going to Make This More Important, Not Less

This is where the current technology conversation gets interesting.

Finance teams are being pushed toward AI-assisted forecasting, continuous planning, and much faster scenario analysis. McKinsey argues that AI can make continuous financial planning practical at scale, while BCG describes planning and forecasting as one of the more developed finance use cases for agentic AI. McKinsey & Company

Wonderful.

Let the machine calculate.

Let it identify relationships.

Let it test thousands of combinations.

Let it surface scenarios a human analyst might not have considered.

I want all of that.

But somebody still has to decide:

Which scenario matters enough to talk about?

Which assumption doesn’t pass the smell test?

Which risk can we actually do something about?

Which signal should management watch?

What decision changes if this happens?

Generating possibilities is becoming cheaper.

Choosing what matters isn’t.

That may end up being one of the more valuable FP&A skills in an AI-heavy Finance function.

Go Look at Your Scenario Model

Seriously.

Open it.

Ignore the outputs for a minute.

Pick your downside case.

Now ask yourself:

Can I explain what actually happened in the business to create this scenario?

Can I identify the operating drivers that changed?

Do I know which early signals would tell me it’s starting?

Do I know what management would do differently?

Do I know when we’d do it?

If you can’t answer those questions, don’t add another scenario.

You have enough.

You may actually have too many.

I don’t want Finance to predict every possible future.

That sounds exhausting, and frankly the future has never shown much interest in cooperating.

I want Finance to help management recognize the few futures that matter early enough to do something about them.

That’s a much harder job than changing a dropdown from Base to Downside.

Unfortunately, it’s also the useful one.

Related Planning Guides

Scenario work gets stronger when it starts with a clear forecasting process and explicit business drivers. If the planning horizon itself is the issue, see when a rolling forecast makes sense.