Stop Pretending the Forecast Is One Number

Every forecast eventually becomes a number.

$15.2 million of revenue. $4.1 million of fundraising. $900,000 of operating surplus.

The precision feels reassuring. It is also frequently misleading.

Because management rarely knows that revenue will be exactly $15.2 million. What management actually believes is something closer to: “There is a reasonable chance we’ll land around $15 million, a meaningful possibility we’ll fall below it, and some upside if several uncertain items break our way.”

That is a distribution. Most financial reports convert that uncertainty into a single number.

The problem with the single-point forecast

Imagine an organization has ten major revenue opportunities. Some are nearly certain. Some are probable. Some are genuine long shots.

Adding the expected value of all ten may produce a forecast of $5 million. The board sees: Forecast revenue: $5 million.

What disappears is the risk surrounding that number.

There may be only a 50% chance of achieving $5 million. There may be a meaningful possibility of receiving $4 million. There may also be upside to $6 million.

Those possibilities matter because organizations make decisions before the final revenue outcome is known. They hire people. Sign contracts. Launch programs. Approve capital spending. Spend cash.

A forecast should therefore help management understand risk, not simply provide a number to enter into a spreadsheet.

Think in confidence levels

One approach is to present several outcomes.

P20 — downside planning case

For this framework, think of P20 as a conservative outcome: a result with substantial probability of being exceeded. It can be useful for testing how the organization performs when several uncertain items do not materialize.

P50 — central case

P50 is the median outcome: roughly equal probability of finishing above or below it if the model is well calibrated.

P80 — upside planning case

In this framing, P80 represents a stronger outcome that requires more favorable results across the portfolio of assumptions and opportunities.

The naming convention matters less than being explicit about what the probabilities mean in your model. Different organizations and software packages sometimes label percentiles differently, so define the convention once and use it consistently.

Instead of asking “What is the forecast?” leadership can ask “What level of financial risk are we willing to operate against?”

Forecast and budget should do different jobs

One of the most damaging habits in financial planning is forcing the forecast to resemble the budget.

A budget expresses intention. A forecast expresses expectation.

Suppose the fundraising goal is $10 million. Halfway through the year, available information suggests the most probable outcome is $8 million.

The budget should still show the $10 million goal. The forecast should show $8 million.

Changing the forecast to $9.5 million because leadership “still wants to push the team” destroys the purpose of forecasting.

Motivation belongs in management. Probability belongs in forecasting.

Use different information for different purposes

Historical behavior

What has this revenue source actually produced over time? A donor who has consistently contributed $100,000 should not automatically be forecast at $500,000 because someone plans to ask for that amount.

Current information

Has the customer renewed? Has a donor meeting occurred? Has the grant moved into final review? Has attendance weakened? The forecast should change when the evidence changes.

Human judgment

Statistical models are not omniscient. A salesperson, fundraiser, program leader, or executive may possess information the historical record cannot see. Human judgment belongs in the process. It simply should not be allowed to become invisible.

Stage probability

Different opportunities have different levels of maturity. A signed agreement should carry a different weight from an introductory meeting. That sounds obvious. Many forecasts still fail to reflect it consistently.

What the board should see

The board does not need a probability model containing hundreds of rows. It needs the implications.

• P20 revenue: $13.8 million
• P50 revenue: $15.0 million
• P80 revenue: $16.1 million

Then connect those outcomes to decisions.

• At the downside case, pause selected hiring and discretionary commitments.
• At the central case, continue according to plan.
• At the upside case, rebuild reserves or accelerate selected investments.

Now forecasting becomes a management tool.

Accuracy is not the only goal

Forecasting teams often measure themselves by how close the final number came to actual results. That matters. But an excellent forecast should also help management make better decisions before the outcome is known.

A forecast that says $15 million and ultimately produces $15 million is accurate.

A forecast that says, “We expect approximately $15 million, but our downside scenario would put revenue near $13.8 million, so we recommend delaying $750,000 of discretionary commitments” may be far more useful.

The second forecast recognizes that uncertainty itself has financial value.

The point of forecasting

Finance cannot eliminate uncertainty. It can make uncertainty visible.

And once uncertainty becomes visible, leaders can decide how much risk they are willing to carry.

That is why the best forecast is rarely one number.

It is a decision framework.

Related reading: forecast scenario planner; five assumptions CFOs should reforecast after the Fed hike; and the board finance dashboard template.