Most expense forecasts fail in the same way: someone projects revenue carefully and then estimates expenses with a flat percentage or a gut feel. The result is a forecast that looks reasonable until reality arrives and the variable costs scale faster than expected. The discipline is mechanical, but it takes care.

Split first: fixed vs variable

The foundation of expense forecasting is separating costs by how they behave with revenue:

  • Fixed costs - costs that stay the same regardless of revenue. Rent, salaries, software subscriptions, insurance, debt service.
  • Variable costs - costs that scale with revenue or volume. Cost of goods, payment processing fees, contractor work, shipping, hourly labor.
  • Semi-variable - costs that have a fixed component plus a variable component. Utilities (base + usage), some software subscriptions (base + per-user), customer support (team + volume).

Each behaves differently and gets forecasted differently. See Fixed Costs vs Variable Costs for the deeper categorization.

Forecasting fixed costs

The easiest part. Take current monthly run rate, add known changes:

  • New hires (start date + loaded cost)
  • Salary increases (effective date + amount)
  • Lease escalations (contract terms)
  • Software renewals (often at higher rates)
  • Insurance renewals
  • Scheduled equipment purchases (depreciation starting)

Most fixed cost forecasts are accurate to within ±5% over 12 months because the inputs are knowable.

Forecasting variable costs

Express each variable cost as a percentage of revenue or as a per-unit cost. Then forecast revenue first, apply the ratios.

Examples:

  • Cost of goods sold: 35% of revenue (from last 12 months average)
  • Payment processing: 2.9% + $0.30 per transaction; or ~3% of revenue
  • Shipping out: $8 average per order × order count
  • Customer support contractor hours: scales with active customer count
  • Sales commissions: 10% of new revenue

Revisit the ratios quarterly. A cost-of-goods ratio that drifts from 35% to 40% over six months is one of the most reliable early-warning signs of margin compression - and catches it in the forecast before it shows up in actual monthly results.

Don't forget lumpy expenses

The category of expenses that wreck unforecasted cash flow:

  • Quarterly estimated taxes
  • Annual insurance renewals (often $5-30K lump sum)
  • Annual software renewals (sometimes 12-15 months in one bill)
  • Annual accounting/audit fees
  • Equipment purchases (capital expenditure)
  • Legal fees on specific matters
  • Conference and event spend

None of these appear in monthly run rate. All of them appear in cash flow when they hit. Build them into the forecast as discrete monthly amounts in the right months - or as monthly accruals if the cash impact is what you're planning.

How granular should the forecast be

Group by category that matters for decisions. 8-12 categories is usually enough:

  • Payroll (fixed)
  • Contractor / freelance (variable)
  • Rent & utilities (fixed)
  • Software & tools (mostly fixed)
  • Marketing & advertising (variable)
  • Cost of goods (variable)
  • Professional services (semi-variable)
  • Insurance (fixed, lumpy)
  • Travel & entertainment (variable)
  • Office & supplies (mostly fixed)
  • Taxes (fixed, lumpy)
  • Other

More granular than that and the forecast becomes a maintenance burden without adding decision-relevant information.

Common expense forecasting mistakes

1. Treating all expenses as fixed

Already covered. The most common error - and the one that makes forecasts most wrong when revenue moves.

2. Forecasting expenses from last month, not the run rate

Any single month has noise. Use a 3-month rolling average for "normal" expense levels.

3. Forgetting payroll true-ups and benefits adjustments

Bonus accruals, year-end true-ups, benefit cost changes. Easy to miss; meaningful when missed.

4. Ignoring inflation

For long-horizon forecasts, fixed costs aren't actually fixed. Rent escalates. Software prices rise. Insurance renews higher. Add 3-5% per year to the fixed cost base for multi-year forecasts.

Related concepts