Cost Forecasting Methods
Executive Summary
Key Takeaways
- ✓ Cost forecasting should begin with classifying each cost line as fixed or variable, since that classification determines which forecasting method is appropriate for it.
- ✓ The percent-of-revenue method is simple and transparent but only appropriate for costs that genuinely scale in proportion to revenue.
- ✓ Driver-based opex build-up ties each cost line to the specific operational driver that actually causes it to move, such as headcount or square footage, rather than to revenue by default.
- ✓ Cost of goods sold construction builds the direct costs attributable to production or service delivery from their own unit-level drivers, typically volume and unit cost.
- ✓ A bottom-up, unit-driven revenue forecast is usually paired with a correspondingly driver-based cost forecast, so that revenue and cost move on a consistent operational basis.
Overview¶
Operating costs, unlike revenue, cannot be forecast reliably using a single method applied uniformly across every line. Different cost lines behave differently as a business scales — some are genuinely fixed within a relevant range, some move in close proportion to revenue, and others move with an entirely different operational driver, such as headcount or production volume. This guide sets out the classification step that should precede any cost forecast, followed by the three principal construction methods used in institutional financial models: the percent-of-revenue method, driver-based opex build-up, and cost of goods sold (COGS) construction. It is the companion guide to Revenue Forecasting Methods, covering the cost side of the same forecast, and relates to the broader methodology framework set out in Forecast Methodologies Overview.
Fixed vs Variable Cost Classification¶
Before selecting a forecasting method, every cost line should be classified according to how it actually behaves as the level of business activity changes.
Fixed costs do not change with the level of activity within a relevant range — a base office lease, a fixed salary, insurance, or a software licence fee typically fall into this category. A fixed cost is generally forecast by holding it flat, or by applying an inflation or contractual escalation assumption, rather than by tying it to a revenue or volume driver.
Variable costs move with the level of activity — a sales commission, a direct material input, or a per-transaction processing fee. A variable cost is forecast by referencing whichever driver actually causes it to move, which is not always revenue.
Semi-variable (mixed) costs contain both a fixed and a variable component — a utility bill with a fixed base charge plus a usage-based component, or a salary structure with a fixed base plus a variable, activity-linked bonus. A semi-variable cost is typically forecast by splitting it into its fixed and variable components and forecasting each separately, rather than treating the blended total as either purely fixed or purely variable.
Total Forecast Cost = Fixed Component + (Variable Rate × Activity Driver)
This classification step matters because it determines which of the three methods below is appropriate for each line — applying a single method to every cost regardless of its actual behaviour is the most common structural weakness in a cost forecast, addressed in the Common Errors section below.
Percent-of-Revenue Method¶
The percent-of-revenue method forecasts a cost line as a fixed percentage of forecast revenue in the same period.
Forecast Cost = Forecast Revenue × Cost as % of Revenue
When to use it. This method is appropriate for a cost that genuinely scales in close proportion to revenue — a payment-processing fee charged as a percentage of transaction value, a royalty or licence fee calculated as a percentage of sales, or a sales commission structured as a fixed percentage of the sale. It is simple, transparent, and easy for a reader to sanity-check against a historical margin.
Limitations. The method's simplicity is also its principal risk: it embeds an implicit assumption that the cost genuinely scales with revenue at a constant ratio, which is frequently not true of costs that are actually fixed, staged, or driven by a different operational variable. Applying a percent-of-revenue assumption to a cost that does not truly behave that way — most commonly headcount-driven salary cost, or a fixed facilities lease — produces a forecast that mechanically tracks revenue rather than reflecting how the cost will actually move.
Driver-Based Opex Build-Up¶
Driver-based opex build-up forecasts a cost line from the specific operational driver that actually causes it to move, rather than defaulting to revenue.
Forecast Cost = Operational Driver × Cost per Unit of Driver
The most common application is a headcount-driven salary build, in which the forecast is constructed from a headcount schedule (a driver representing planned hires and departures by period) multiplied by an average salary or fully loaded cost-per-employee assumption:
Forecast Salary Cost = Forecast Headcount × Average Fully Loaded Cost per Employee
Other common opex drivers include square footage for facilities cost, transaction count for a per-transaction processing cost that is not itself revenue-proportional, and unit shipped for a logistics or fulfilment cost.
When to use it. Driver-based opex build-up is appropriate for any cost that is more accurately explained by an operational variable other than revenue, which in practice covers most of a typical operating cost base beyond directly revenue-proportional fees. It is also the method that makes a cost forecast most directly auditable, since each cost line traces to an explicit, labelled driver rather than an assumed ratio.
Limitations. Driver-based build-up requires the underlying operational driver — a headcount plan, a square-footage schedule — to itself be forecast with a defensible basis. A driver-based cost forecast built on an unsupported headcount plan is no more reliable than a percent-of-revenue assumption; the method shifts the burden of justification onto the driver itself rather than eliminating it.
Cost of Goods Sold (COGS) Construction¶
Cost of goods sold, the direct costs attributable to producing the goods or delivering the services sold in a period, is typically built up from unit-level drivers, mirroring the bottom-up construction used on the revenue side.
Forecast COGS = Forecast Unit Volume × Forecast Unit Cost
Where a business sells multiple product lines with different unit economics, COGS is generally forecast separately by product line and then aggregated, rather than as a single blended unit-cost assumption:
Forecast COGS = Σ [ Unit Volume(i) × Unit Cost(i) ] for each product line i
When to use it. COGS construction from unit volume and unit cost is the standard method wherever a business has an identifiable unit of production or service delivery, and is typically paired with a bottom-up, unit-driven revenue forecast so that gross margin is derived consistently from the same underlying volume driver on both sides, rather than revenue being built bottom-up while COGS is forecast as an unrelated percentage of that revenue.
Limitations. Unit cost itself is frequently not constant across the forecast period — input cost inflation, supplier renegotiation, or scale-driven unit-cost reduction can all cause the unit cost assumption to change period over period, and a COGS build that holds unit cost flat across a multi-year forecast without documenting that assumption understates this source of uncertainty.
Choosing and Combining Methods¶
| Method | Best suited to | Data requirement |
|---|---|---|
| Fixed cost forecast (held flat or escalated) | Costs that do not change with activity level within a relevant range | Current cost base, escalation or inflation assumption |
| Percent-of-revenue | Costs that scale in close proportion to revenue (processing fees, royalties, commissions) | Historical cost-to-revenue ratio |
| Driver-based opex build-up | Costs driven by an operational variable other than revenue (headcount, square footage) | A defensible forecast of the underlying driver |
| Cost of goods sold construction | Direct production or service-delivery costs where a unit of output exists | Forecast unit volume and forecast unit cost |
A single institutional cost forecast typically applies all four approaches simultaneously across different lines of the same cost schedule — fixed costs held flat or escalated, revenue-proportional fees forecast as a percentage of revenue, headcount-driven costs built from a headcount schedule, and COGS built from unit volume and unit cost. Applying one method uniformly across the entire cost base, most commonly defaulting every line to percent-of-revenue for simplicity, is a common source of forecast unreliability addressed below.
Common Errors¶
| Error | Description | Risk |
|---|---|---|
| Every cost line forecast as a percentage of revenue | No classification step performed before selecting a method | Fixed and driver-based costs mechanically track revenue instead of their actual behaviour |
| Fixed cost forecast with no escalation assumption | A cost held flat with no stated inflation or contractual escalation basis | Forecast understates cost in later periods without disclosing that assumption |
| Headcount-driven cost forecast with no headcount schedule behind it | Salary cost line references a percentage-of-revenue assumption instead of a driver | Cost forecast is not auditable against the actual hiring plan |
| Unit cost held flat across a multi-year COGS forecast | No input-cost inflation or scale-efficiency assumption documented | Understates a material source of margin uncertainty |
| Semi-variable cost treated as purely fixed or purely variable | Mixed cost not split into its fixed and variable components | Forecast cost does not respond correctly to a change in the activity driver |
Best Practices¶
Classify every cost line as fixed, variable, or semi-variable before selecting a forecasting method, and document the classification alongside the assumption itself. Tie each variable or semi-variable cost to the operational driver that actually causes it to move, rather than defaulting to revenue for simplicity. Where percent-of-revenue is used, confirm the underlying relationship is genuinely proportional and state the historical basis for the assumed ratio. As with revenue, every driving assumption — a cost ratio, a headcount plan, a unit cost — should be entered once on a labelled assumptions tab and referenced by the cost formula, as set out in Assumption Design Best Practices.
Continue Reading¶
Prerequisites¶
- Financial Forecasting in Financial Models — the parent pillar
Related Technical Guides¶
- Revenue Forecasting Methods — the companion guide covering the revenue side of the same forecast
- Forecast Methodologies Overview
- Assumption Design Best Practices
Related Glossary¶
Related Checklists¶
How OXXON tests thisRun a free structural check with FMAE
Frequently Asked Questions
What is the first step in forecasting operating costs?
Classifying each cost line as fixed or variable. A fixed cost does not change with the level of business activity within a relevant range, such as base rent or a fixed salary. A variable cost moves with activity, such as a sales commission or a direct material cost. This classification determines which forecasting method is appropriate for each line, addressed in the Fixed vs Variable Cost Classification section below.
When is the percent-of-revenue method appropriate for forecasting a cost?
When a cost line genuinely scales in close proportion to revenue, such as a payment-processing fee charged as a percentage of sales, or a royalty. It is not appropriate for a cost that is fixed, or that scales with a different driver than revenue, such as headcount-driven salary cost in a business growing revenue per employee rather than by adding headcount in proportion to sales.
What is driver-based opex build-up?
A method that forecasts a cost line from the specific operational driver that actually causes it to move, rather than from revenue by default — for example, forecasting total salary cost from a headcount schedule and an average salary assumption, or forecasting facilities cost from a square-footage driver and a cost-per-square-foot assumption.
How is cost of goods sold forecast?
Typically built up from unit-level drivers, most commonly forecast unit volume multiplied by a forecast unit cost, mirroring the bottom-up construction used on the revenue side, described in the Cost of Goods Sold Construction section below.
Should every cost line use the same forecasting method?
No. Different cost lines behave differently as a business scales, and applying a single method, most commonly percent-of-revenue, to every cost line regardless of its actual behaviour is one of the most common structural weaknesses in a cost forecast, addressed in the Common Errors section below.
How does cost forecasting relate to revenue forecasting?
The two are companion disciplines within the same forecast. A bottom-up, unit-driven revenue build is typically paired with a correspondingly driver-based cost build, for example variable costs tied to the same unit-volume driver used on the revenue side, so that revenue and costs move on a consistent operational basis rather than costs being forecast in isolation as a fixed percentage of whatever revenue figure results, described on the companion Revenue Forecasting Methods guide.
What is the risk of forecasting every cost as a percentage of revenue regardless of its actual behaviour?
It embeds an implicit assumption that every cost line scales with revenue at a constant ratio, which understates cost in a period of unusually fast revenue growth and overstates it in a period of slow growth, since real operating costs generally lag or lead revenue rather than moving in lockstep with it.
Related Articles
Financial Forecasting in Financial Models
Financial forecasting is the process of projecting a business's future financial performance from a defined set of operating drivers and assumptions, structured so that every forecast line traces back to a labelled, auditable input rather than a value typed directly into a calculation. It underpins every model built for valuation, budgeting, financing, or investment decision-making, and it is also one of the areas of a financial model most prone to silent structural failure, since a forecast that looks complete can still rest on drivers that are hardcoded, undocumented, or inconsistently applied from one period to the next. This page is the hub for the Knowledge Centre's forecasting content: what a forecast driver is, the major forecasting methodologies and when each applies, the governance distinction between a budget and a forecast, rolling forecasts, and how forecasting failure modes map onto FMAE's existing structural audit rule taxonomy.
Revenue Forecasting Methods
Revenue can be forecast using several structurally different methods, and the choice of method has a direct effect on how defensible and auditable the resulting forecast is. This guide sets out the four principal methods used in institutional financial models — top-down forecasting from market size and share, bottom-up forecasting from unit economics, trend and growth-rate extrapolation from historical results, and cohort-based forecasting for subscription and other recurring-revenue businesses — with guidance on when each method is appropriate and how the methods can be combined within a single forecast.
Forecast Methodologies Overview
Before a forecast line is built, a methodology has to be chosen for how it will be projected. This guide compares the four principal forecasting methodologies used across a financial model's revenue and cost lines: top-down forecasting, which starts from a macro or market-level figure and works down; bottom-up forecasting, which builds up from granular unit economics; driver-based forecasting, which structures the model around a defined set of operating drivers regardless of direction; and the percent-of-sales method, which forecasts a line as a constant ratio of revenue. It sets out how the four relate to each other, when each is most defensible, and how they are applied to the revenue and cost sides of a forecast.
Forecast Driver
A forecast driver is a labelled input cell, most commonly a growth rate, a margin percentage, a unit count, or a price, that a forecast formula references rather than embeds directly. It is the structural unit that makes a forecast auditable and sensitizable, because changing the driver cell changes every downstream calculation that depends on it, consistently and traceably. A forecast driver is structurally distinct from a hardcode, a value typed directly into a calculation cell with no traceable source, even where the two produce an identical output in a given period.
Assumption Design Best Practices
How a forecast's assumptions are designed determines whether the forecast can actually be audited, sensitized, and defended in front of a reviewer, independent of whether the assumed values themselves are reasonable. This guide sets out five construction disciplines for assumption design: separating input cells from calculation formulas, labelling every assumption clearly with its unit, consolidating assumptions onto a dedicated tab, entering each driver once at a single point rather than repeating it, and structuring input cells so they can be sensitized cleanly without breaking the calculations that depend on them.