A forecast is not a prediction of the future. It is a structured argument about what you believe is going to happen—and why.
There is a comforting moment in every forecasting process when the spreadsheet finally balances.
Revenue: $48.7 million.
EBITDA: $4.2 million.
Ending cash: $6.1 million.
The formulas work. The formatting is immaculate. The totals reconcile.
Everyone can go home.
Except for one small problem.
None of those numbers knows what is going to happen.
The customer doesn’t know they’re supposed to sign on October 1.
The implementation team hasn’t seen the hiring plan.
Accounts receivable did not receive the memo about your 45-day collection assumption.
And the spreadsheet, despite looking extremely confident, has no idea either.
This is not a forecasting problem.
It’s an assumptions problem.
The number is the least interesting part of the forecast
Executives naturally focus on outputs.
What’s revenue going to be?
Where will EBITDA land?
How much cash will we have?
Those are important questions. But the answers are simply the mathematical consequences of everything management assumed before Excel produced them.
Revenue will be $48.7 million if the pipeline converts at the expected rate.
EBITDA will be $4.2 million if hiring happens when planned, margins hold and the organization realizes the expected efficiencies.
Ending cash will be $6.1 million if customers pay when expected, expenses land when expected and nothing expensive decides to become interesting.
That’s a lot of ifs hiding behind three very precise numbers.
A useful forecast makes those assumptions visible.
A dangerous one hides them behind decimal places.
Precision and accuracy are not the same thing
There is something wonderfully reassuring about a forecast that says revenue will be $48,734,219.
Look at all those digits.
Surely someone knows what they’re doing.
But adding precision to an uncertain assumption does not make the assumption more accurate.
If the forecast depends on a $3 million customer launching sometime in Q4, the important question probably isn’t whether October revenue should be $812,417 or $814,092.
It’s:
How confident are we that the customer launches in October at all?
This is where forecasts often become unintentionally misleading.
The model is mathematically precise.
The business isn’t.
And management starts debating the fourth decimal place of an assumption that could be wrong by 30%.
Every forecast has a few assumptions doing most of the work
A 20-tab financial model can create the impression that hundreds of variables are equally important.
They aren’t.
Usually, a handful of assumptions explain most of the difference between the forecast working and not working.
Maybe it’s:
- when two large customers launch;
- whether utilization reaches the expected level;
- whether a pricing change sticks;
- how quickly open positions are filled;
- whether gross margin improves;
- when a large receivable actually converts to cash.
Those are the assumptions leadership should know cold.
Not because everything else is irrelevant.
Because those are the things that can change the answer.
What five assumptions would I be most worried about if I were betting my own money on this number?
Those belong in the conversation.
The forecast should tell you what has to be true
One of the simplest ways to improve forecasting is to stop asking:
“Do we believe the forecast?”
That’s almost impossible to answer.
Ask instead:
What has to be true for this forecast to happen?
Now you have something useful.
Customer A must launch by October 1.
Open headcount must stay below 12 positions through Q3.
Gross margin must improve from 31% to 34%.
Receivables over 90 days must decline by $1.5 million.
The new service line must reach 2,000 monthly units by December.
Those are observable conditions.
And once you’ve identified them, you can monitor them.
That’s where a forecast stops being a finance exercise and starts becoming a management tool.
Stop quietly replacing bad assumptions
Here’s a familiar forecasting ritual.
Last month, management assumed the new customer would launch July 1.
July arrives.
No launch.
The forecast is updated.
Now the customer launches August 1.
August arrives.
Still no launch.
September 1 it is.
At no point does anyone explicitly say:
Our assumption has been wrong three months in a row.
The date simply moves.
This is how forecasts preserve the appearance of reasonableness while slowly becoming fiction.
When a major assumption misses, don’t just update it.
Keep score.
What did we assume?
What actually happened?
Why were we wrong?
Does that tell us anything about the assumptions we’re making now?
A forecast should learn.
A miss is information
Organizations sometimes treat forecast misses like grades.
Beat forecast: good.
Miss forecast: bad.
But a forecast isn’t a performance target, even though the two are often intertwined.
If revenue misses because sales execution deteriorated, that’s a performance issue.
If revenue misses because management consistently overestimates implementation speed, that’s an assumption issue.
If cash misses because a payer historically takes 75 days to pay and the model keeps assuming 45, that’s not bad luck.
That’s a hint.
The purpose of comparing actuals to forecast isn’t just to explain the variance.
It’s to improve the next forecast.
Otherwise, you’re performing an elaborate monthly ritual in which everyone explains why the last number was wrong before producing a new number using essentially the same logic.
Scenario planning is not pessimism
There is sometimes resistance to downside scenarios because they feel unnecessarily negative.
We don’t want to manage to the downside.
Good.
Don’t.
But you should probably know what it looks like.
A base forecast answers:
What do we currently expect to happen?
A downside scenario asks:
What happens if one or two important assumptions don’t?
A useful downside scenario isn’t:
Revenue mysteriously declines 20%.
It’s:
Customer A launches 90 days late, hiring continues as planned and collections stretch by 15 days.
Now the scenario tells you something operationally useful.
How much cash do we lose?
When?
What decisions would we make differently?
How much time do we have to make them?
That’s not pessimism.
That’s knowing where the exits are before you need one.
Cash has a particularly good sense of humor
An income statement can tolerate a surprising amount of optimism.
Cash is less accommodating.
Revenue may be recognized.
EBITDA may look healthy.
The customer may even agree that they owe you the money.
Your bank account remains stubbornly interested in whether they actually paid.
That’s why short-term cash forecasting deserves a different level of specificity.
When liquidity matters, “A/R collections” isn’t an assumption.
Which receivable?
How much?
When?
Why do we believe that?
What happens if it slips one week?
Two?
Four?
A 13-week cash flow forecast is particularly good at exposing assumptions management has been able to avoid confronting in a longer-range model.
Which is precisely why it can occasionally ruin an otherwise pleasant Tuesday.
Forecast ranges are sometimes more honest than forecast points
Executives often want one number.
Boards often want one number.
Banks definitely enjoy one number.
Reality remains uncooperative.
If the organization genuinely cannot estimate an outcome more precisely than a reasonable range, pretending otherwise doesn’t improve the analysis.
Sometimes:
$46 million–$49 million
is more useful than:
$47.6 million
The question is what creates the range.
If the difference is primarily one customer launch, say that.
If it’s utilization, say that.
If it’s a contract negotiation, say that.
Uncertainty isn’t the problem.
Unexamined uncertainty is.
Your forecast should change decisions
A forecast that doesn’t change anything is mostly a reporting artifact.
A useful forecast helps management decide:
Do we hire now or wait?
Do we accelerate an investment?
Do we reduce spending?
Do we need financing?
Can we make the acquisition?
Should we change pricing?
When do we need to intervene?
This is why the best forecasting conversations don’t end with:
“So that’s the forecast.”
They end with:
Given the forecast, here’s what we’re going to do.
That is the point.
Build the forecast around the questions
A better forecasting process doesn’t necessarily require a more complicated model.
It requires better questions.
- What are we assuming?
- Which assumptions matter most?
- What evidence supports them?
- What has changed since the last forecast?
- Where have our assumptions historically been wrong?
- What would tell us early that we’re wrong this time?
- What happens if we are?
- And what decision would that change?
Answer those well and something useful happens.
The forecast becomes less about predicting the future.
And more about being ready for it.
The spreadsheet can keep the decimal places.
Just don’t confuse them with certainty.
13-Week Cash Flow Forecast
A practical cash forecasting model designed to help organizations understand near-term liquidity, make timing assumptions explicit, and see pressure developing before the bank balance delivers the news personally.