There is a one-page format that, if installed as the standard for every forecast that reaches the CFO's desk, would change a company more than any planning software, BI rebuild, or FP&A reorganization we have seen attempted.
The format is five lines. It can be written on an index card. It does not require a new tool, a new hire, or a new model. What it requires is the willingness to ask, of every number that gets presented as a forecast, the same five questions — and to send the document back if any of them are missing.
We have argued the underlying philosophy elsewhere: a forecast that is a single number is not a forecast. This essay is narrower. It is about the format. The shape of the artifact. The thing the CFO should be able to point at and say: this is what a forecast looks like in our company; if it doesn't look like this, it isn't one.
The five lines
Here is the format, in full.
Line 1. The 80% confidence interval. Two numbers: a low and a high. The producer of the forecast is claiming that there is roughly an 80% chance the actual outcome lands between them. The interval should be derived honestly — meaning, derived from the forecaster's actual belief about the distribution of outcomes, not negotiated backward from a number the room is hoping to hear.
Line 2. The point estimate, if one is required. This is the number that goes into the plan, the guidance, or the model. It should sit clearly inside the interval, and it should be labelled as what it is — a chosen point, not a forecast.
Line 3. The two or three named drivers that would move the interval. Not "market conditions." Specific variables, with directions. "If our second-half net retention holds above 110%, we are at the upper half of the range; if it slips below 105%, we are at the lower half." A reader who watches the named drivers should be able to update the forecast in real time as the world delivers new data.
Line 4. The historical track record on forecasts of this type. How often has this team, on this kind of forecast, had its 80% intervals contain the actual? If the answer is "we don't know," the line should say so. The act of writing "we don't know" is, by itself, half the value of the format.
Line 5. What would update the range. A single paragraph. What evidence, between now and the next review, would cause the producer to widen, narrow, or shift the interval? Without this line, the forecast is a static document; with it, the forecast is a living view that the team has explicit instructions for how to revise.
That is the entire format. Five lines, all of them short, all of them answerable, none of them optional.
The format most companies actually use
Compare the format above to what most forecasts on a CFO's desk actually look like.
The typical artifact is a single number, sourced from a model the CFO cannot interrogate in the meeting, supported by a paragraph of qualitative justification, with no historical context for how forecasts from this team have performed and no statement of what would cause the producer to revise. The number is presented with the confident affect of someone who has done the work. The work has indeed been done. It is just not the work that would help the CFO decide anything.
We have sat through hundreds of these presentations. The pattern is almost invariant. The model is sophisticated. The number is precise to several significant figures. The conversation in the room is about whether the number is "right," which is a question nobody can answer until the period is over. The conversation that would help — how wrong might this be, and in which direction — is the conversation the format makes impossible.
Most forecast meetings are exercises in defending a point estimate. The five-line format converts them into exercises in interrogating a range. The first kind of meeting produces alignment. The second kind produces decisions.
There is also a quieter cost. A single-number forecast, presented in the typical format, gives the room no language for disagreement. To disagree is to produce a different single number, which the presenter will then defend as theirs has been defended. The format becomes a contest of conviction rather than a conversation about uncertainty. The room that pushes hardest wins. The actual underlying truth, which is that nobody knows the answer with any precision, is the casualty.
Why this is a five-line change with outsized consequences
The cultural shift the format produces is larger than the format itself, and it is worth being explicit about why.
The first consequence is that uncertainty becomes legible. In a single-number culture, the uncertainty exists — every forecaster knows their number is wrong — but it lives in private conversations, hallway hedges, and unrecorded skepticism. In a range culture, the uncertainty is on the page, in writing, where it can be reasoned about. The company has not become more uncertain. It has become honest about the uncertainty it always had.
The second consequence is that the conversation moves from the number to the drivers. A CFO reviewing a five-line forecast does not spend the meeting asking whether the point estimate is right. They spend it asking about the named drivers. Are those the right drivers? Are there others? What is the sensitivity to each? This is the conversation that would help the company prepare for the range of outcomes, and it is the conversation the standard format prevents.
The third consequence is that track record becomes data. Line four of the format — the historical accuracy of forecasts of this type — has to come from somewhere. The first time a team has to fill it in, the answer is usually "we have not been tracking that." The second year, with the habit installed, the answer is a number. By the third year, the company has a calibration record for every forecaster on every type of forecast, and the question of which forecasters to trust on which questions has an answer. This is, quietly, one of the most valuable institutional assets a finance organization can build, and the five-line format is what generates it.
The fourth consequence is that the format compounds with reforecasting. We argued in our piece on annual planning that the plan should be a living view, refreshed on a regular cadence. The five-line format is what makes that refresh tractable. Line five — what would update the range — gives the next review a clear agenda. The team comes back, reports against the named drivers, updates the interval, and the cycle continues. Without the format, the refresh is a blank-page exercise; with it, the refresh is a structured revision.
Objections worth addressing
There are three objections we hear consistently, and each is worth taking seriously.
The first is operational overhead. "We produce hundreds of forecasts a year; we can't do this for all of them." Correct. The format is for the forecasts that matter — the ones that drive material capital, hiring, or strategic decisions. For most companies, this is a list of ten to thirty per year. The discipline is in deciding which forecasts get the treatment and being firm that those forecasts do not move forward without it.
The second is false precision. "An 80% interval is just two made-up numbers instead of one." Partially true. The first interval a forecaster produces, before they have a track record, is roughly as well-informed as the point estimate. The difference is that the interval generates evidence about whether the forecaster was calibrated, which the point estimate does not. By the second year, the interval is informed by the prior year's hit rate. By the third year, the forecaster's intervals are meaningfully better-calibrated than their old point estimates ever were. The discipline produces its own data.
The third is political. "If I make people put ranges in writing, they will pad them to ensure they always hit." Possible. The defense is the calibration record itself. A forecaster whose 80% intervals are containing the truth 99% of the time is producing useless ranges, and the record will show it. The format is self-policing in a way the single-number format never is — because in a single-number culture, there is no record of how wrong the number was relative to what was claimed.
How to install it
The mechanics, for a CFO who wants this to be the format by next quarter:
Pick the ten forecasts that most matter — usually a mix of revenue, headcount, capex, key product metrics, and any large discretionary spend. Issue the format as the new standard for those ten. Send back, without engaging on substance, any forecast that arrives in the old format. The first few will come back incomplete; this is a one-time cost.
Build a simple sheet — a row per forecast, the five lines, and a column for the actual when it lands — that becomes the calibration record. After two cycles, the record itself becomes the most useful artifact in the planning process; FP&A will start referring to it before producing the next forecast, which is exactly what should happen.
That is the install. It costs nothing. It requires no software. The political cost is real but bounded — the executives whose forecasts will look worst under the new format are usually the ones whose forecasts have been quietly wrong for years, and the format will make that visible. Whether that visibility is welcome depends on whether the CFO and CEO have the appetite for it.
In our experience, the appetite is the limiting reagent. The format is trivial. The willingness to install it — and to mean it when forecasts get sent back — is what separates the companies that do this from the companies that talk about doing it. We help CFOs install the discipline and run it for the first two quarters until it is self-sustaining. The format is, as advertised, five lines. Everything else is whether you actually use them.
Writes about decision quality at Bayeseon. Reach the team at hello@bayeseon.com.