How accurate is the plan?
№ 232 · Forecast vs actuals variance
Definition
MetricForecast error, actual against forecast
Unitpercent error, and percentage points of miss on a growth rate
Bar of variance per period (actuals minus forecast). Pull planning data and actuals from finance system. Consistent miss in specific line items signals weak forecasting in that area. Example: Revenue +4% over plan; S&M +18% over plan; R&D +2% (S&M is the variance leak).
- S&M
- Sales and marketing. All go-to-market cost: salaries, commission, media spend and tooling.
- R&D
- Research and development. Engineering, product and design cost, usually shown as a percent of revenue.
Benchmarks
| Bottom 30% | Median | Top 30% |
|---|---|---|
| above 2.4 percent error for mature organisations, above 20 points of growth-rate miss in venture-stage software | 1.8 percent error in large cross-industry organisations, 8 to 10 points of growth-rate miss in venture-stage software | under 1.5 percent error on a one-quarter revenue forecast for a mature business, under 5 percent for a growth-stage business |
two populations reported separately: APQC Open Standards Benchmarking cross-industry participants for total sales forecast percentage error, and private B2B SaaS annual plan against actual growth rate for FY2024 and FY2025 · This is the card where one global number does the most damage. APQC's cross-industry data has top performers at 1.3 percent total sales forecast error, the median at 1.8 percent and lower-tier organisations at 2.4 percent, but those are large, mature, mostly repeat-revenue businesses and a venture-stage company will never see those figures. The relevant comparison for growth-stage businesses is a different one entirely: Benchmarkit has private SaaS companies planning a median 35 percent growth rate against a median 26 percent actual, a 9-point systematic overshoot that repeats year after year. APQC's demand forecast accuracy series, a median of 85 percent across 1,068 organisations, is a units-based supply chain measure and is not comparable to either. Always state horizon, denominator and whether the error is signed or absolute before comparing anything on this chart.
| Category | Bottom 30% | Median | Top 30% |
|---|---|---|---|
| Large cross-industry organisations, total sales forecast percentage error | 2.4 percent | 1.8 percent | 1.1 to 1.3 percent |
| Private B2B SaaS, annual growth plan against actual | more than 20 points of overshoot | planned 35 percent against 26 percent actual, a 9-point overshoot | within 3 points of plan |
| Cross-industry monthly demand forecast accuracy, units basis | below 75 percent | 85 percent | above 92 percent |
When it looks bad
The variance bars sit on the same side of zero month after month, which is bias rather than error, and each reforecast moves toward the actual instead of the actual moving toward the plan.
Twelve consecutive months of revenue landing 6 to 11 percent below forecast, with every reforecast cutting the remaining year by roughly the amount already missed.
What to do about it
- Track signed bias separately from absolute error and chart both. A team with 8 percent absolute error and near-zero bias has a noise problem; a team with 8 percent error consistently on one side has an incentive problem, and only the second one is fixed by changing who owns the number.
- Move the largest revenue line to driver-based forecasting, projecting customers, price and volume separately and letting revenue fall out. It attributes the miss to a specific driver rather than to the forecast as a whole, which is what makes the next forecast better rather than just later.
- Put the prior forecast on every chart alongside the current one. APQC's top performers run 1.3 percent error against a 1.8 percent median, and the difference is mostly process, specifically that people outside finance contribute to the number and can therefore be held to it.
- Set the plan growth rate against peer actuals, not peer plans. Benchmarkit's finding that private SaaS companies plan 35 percent and deliver 26 percent means the industry overshoots by roughly 9 points as a matter of course, so a plan anchored on other companies' plans inherits their error before it starts.
Sources
- APQC (via CFO.com) Reports top-performing organisations at a 1.3 percent error rate on the total sales forecast, a median of 1.8 percent and lower-tier organisations at 2.4 percent, across APQC cross-industry benchmarking data, and notes that a 2 percent miss on a $20M business is a $400,000 shortfall. cfo.com ↗
- APQC (via CFO.com) Reports top performers at a forecasting error of 1.1 percent or less through APQC's Open Standards Benchmarking in planning and management accounting. cfo.com ↗
- Benchmarkit Reports private SaaS companies planning a median growth rate of 35 percent for 2025 against a median actual growth rate of 26 percent in 2024, with companies in every revenue band planning higher growth than they delivered the prior year. benchmarkit.ai ↗
- Xorosoft (reporting APQC) Reports a median average monthly demand forecast accuracy of 85 percent across 1,068 organisations in APQC's dataset, and cautions that the figure is only comparable against a matching formula, horizon and aggregation level. xorosoft.com ↗
- APQC Defines the Open Standards Benchmarking measure for overall financial forecast accuracy as part of the process efficiency set for planning, budgeting and forecasting. apqc.org ↗
- APQC Publishes APQC planning and management accounting benchmarks including total cost to plan, budget and forecast and cycle time to complete the annual budget, by industry. apqc.org ↗
- SaaStr (reporting SVB) Reports that in 2025 burn increased about 50 percent and revenue growth about 75 percent in the year following a raise, so post-financing periods are systematically the hardest to forecast. saastr.com ↗
- Sapphire Ventures and KeyBanc Capital Markets Reports quota attainment expectations expanding from about 70 percent to about 75 percent across more than 100 private SaaS companies, which is the sales-side input that drives most revenue forecast error. sapphireventures.com ↗