Bottom-up forecasting builds a revenue projection from granular inputs such as pipeline deals, average deal size, win rates and capacity, rather than from a share of a total market. It produces a number tied to operational reality, and one that can be interrogated line by line.
Bottom-up and top-down are answering different questions
The weakness of a top-down number is that nobody can disprove it. Pick a market, pick a share, and no one in the room has grounds to say the share is wrong. A bottom-up number can be shown to be wrong, which is the reason to build one. Every line in it is a deal, a conversion rate or a named person, and each of those can be checked against what happened last quarter.
Both have a place. The top-down number frames the ambition and the bottom-up number tests whether the next twelve months can carry it. The mistake is presenting a top-down number as though it were operational, which is what most boards have learned to look for. ASIC applies a formal version of that test to any forecast published in a disclosure document, where an issuer has to be able to show reasonable grounds for prospective financial information. No board pack is held to that bar, but it is a fair question to put to your own number.
| Top-down | Bottom-up | |
|---|---|---|
| Starting point | Total addressable market, then a share of it | Individual deals, units or customers you can name today |
| Core inputs | Market size, growth rate, assumed share | Pipeline by stage, conversion by stage, average deal size, sales and delivery capacity |
| Failure mode | Unfalsifiable. Nobody can prove the share is wrong | Anchors too hard on recent history and misses a step change in the business |
| Best used for | Framing long-run ambition and market opportunity | The next twelve to eighteen months, hiring plans, and anything cash depends on |
The inputs a bottom-up model needs
For a sales-led business the core build is: open pipeline by stage, conversion rate by stage taken from your own history rather than a benchmark, average deal size by segment, and sales capacity, meaning how many quota-carrying people you actually have and when the newest ones become productive. For a product or unit-led business, substitute unit volume, price and production or delivery capacity.
Two of those inputs are usually missing. Conversion rate by stage requires enough closed history to be a rate rather than an anecdote, and capacity requires an honest ramp assumption for new hires. Both are where a model that looks rigorous stops being rigorous, because both are quietly guessed. If they are guessed, say so on the tab.
Timing is half the value
An annual bottom-up total is not much more useful than a top-down one. The value comes from mapping when revenue is expected to land, month by month, from the stage a deal is in now and how long deals at that stage historically take to close. That is what turns a forecast into something a hiring plan and a cash flow can hang off.
It is also where the difference between revenue and cash shows up. A deal that closes in March and bills quarterly in arrears is a March revenue event and a June cash event. Getting the timing right is usually more consequential than getting the total right, because the total is an argument and the cash date is a fact.
The same discipline applied to the whole revenue base is how a revenue forecast should be assembled.
Running the gap against the top-down target
The most useful output of a bottom-up build is not the number. It is the gap between that number and the target the board has already been given. Quantify the gap, then decompose it: how much is pipeline volume, how much is conversion, how much is deal size, how much is capacity. Each of those has a different fix and a different cost.
This is also the moment a finance function proves what it is for. The job of a capable finance leader in a decision like this is to provide better decision context by outlining trade-offs, reasons for deferral, or which assumptions need confirming, rather than simply blocking.[1] A bottom-up model that arrives as a refusal gets ignored. The same model presented as three costed options gets acted on.
Who can build one, and who can defend it
Building the model is an FP&A skill. When I advise finance professionals on positioning themselves, the experience worth leading with is exactly this: FP&A, financial modelling, budgeting and forecasting, alongside M&A and capital raising, over the routine processing work.[2] That is where the market demand sits, and it is a reasonable proxy for what to look for when you hire.
Defending it is a separate skill and it is the one worth testing for. When I design a take-home task for a finance candidate the purpose is not a picture-perfect answer. It is to see their assumptions, their thinking process, and how they handle the follow-up questions.[3] A bottom-up forecast is a hundred assumptions in a trench coat, so a candidate who cannot narrate their own assumptions under mild pressure will not survive a board meeting.
Forecast accuracy is also measurable, and worth measuring. I encourage candidates to put metrics against achievements, and improving forecast accuracy from 50 percent to 95 percent is the kind of figure that demonstrates real impact.[4] If nobody in your business could quote that number for the last four quarters, the forecast is not being held to account by anyone. Our report on how founders should build a financial model covers the structure side of this in more detail.
If nobody in the business can build this today, the question is what the finance team should look like at your stage.
Common questions
What is bottom-up forecasting?
Bottom-up forecasting builds a revenue projection from granular operational inputs rather than from a share of a market. In a sales-led business that means open pipeline by stage, historical conversion rates by stage, average deal size and sales capacity; a unit-led business substitutes volume, price and delivery capacity.
How is bottom-up forecasting different from top-down?
Top-down starts with a market size and applies an assumed share, which makes it fast and effectively unfalsifiable. A bottom-up build multiplies out from deals or units you can name, so it takes longer and can be tested line by line. Most businesses need both: top-down to frame ambition, bottom-up to test whether the next year can carry it.
What is a bottom-up versus top-down gap analysis?
It is the comparison between the number the bottom-up model produces and the target already given to the board. Decomposing that gap into its causes matters more than the headline: pipeline volume, conversion rate, deal size and capacity. Each has a different cost to close, which turns a bad-news number into a set of costed options.
Who should build a bottom-up forecast?
Building it is an FP&A skill, whether that sits with an FP&A analyst, a management accountant or a head of finance depending on company size. Whether the candidate can defend it under questioning is a different matter, and the one worth testing at interview. A useful test is a take-home task judged not on a perfect answer but on whether the candidate can narrate their assumptions and handle follow-up questions.
References
- The point I make here: provide better decision context by outlining trade-offs, reasons for deferral or necessary assumption confirmations, rather than simply blocking.
- The advice I give on this: highlight FP&A experience including financial modelling, budgeting and forecasting, as well as M&A work and capital raising, ahead of routine tasks, because that is what the market values. Tom Hunter and the finance recruitment specialism behind that view are the subject of an interview on the Honest Wealth Builders podcast.
- From a search I ran: the purpose is not a picture perfect response but to test the candidate's assumptions, thinking process, and how they handle follow-up questions.
- What I coach people to do: it is worth putting metrics against achievements, such as showing forecast accuracy improved from 50% to 95%, to illustrate the level of impact and value a candidate brought to the role.