SaaS Revenue Forecasting Models: Building Accurate Projections
Master revenue forecasting. Build bottom-up models, forecast ARR, and improve forecast accuracy.
Key Takeaways
- Bottom-up ARR forecasting: Start with existing ARR, then model each component. Formula: Next month ARR = Current ARR + New ARR + Expansion ARR - Churned ARR - Contraction ARR. Example: £1M ARR + £80K new + £30K expansion - £20K churn - £5K contraction = £1.085M. Project each component monthly using pipeline data (new), historical rates (expansion/churn). Accuracy: Bottom-up is 85-95% accurate for 3-month forecasts.
- Pipeline-based forecasting: Weight pipeline by stage probability. Stages: Discovery (10%), Qualified (25%), Proposal (50%), Negotiation (75%), Verbal commit (90%). Example: £500K pipeline at Discovery = £50K weighted, £200K at Proposal = £100K weighted. Total weighted pipeline: £150K expected new ARR. Compare weighted pipeline to quota (pipeline coverage ratio target: 3-4x quota). If quota is £100K and pipeline is £300K weighted = 3x coverage (adequate).
- Forecast accuracy measurement: Track forecast vs actuals monthly. Metrics: (1) Mean Absolute Percentage Error (MAPE) — target <15%, (2) Forecast bias (consistently over/under), (3) Weighted pipeline accuracy by stage. Example: Forecasted £100K new ARR, actual £92K = 8% error (good). Track by sales rep (identify optimists/pessimists). Calibrate stage probabilities quarterly based on historical conversion rates.
Building Accurate SaaS Revenue Forecasting Models
Creating reliable revenue projections for planning and investor reporting. **Bottom-up ARR forecast model** The ARR bridge framework: Opening ARR + New ARR (new logos) + Expansion ARR (upsells, cross-sells) - Churned ARR (lost customers) - Contraction ARR (downgrades) = Closing ARR Monthly forecast example: | Component | Jan | Feb | Mar | Q1 total | |---|---|---|---|---| | Opening ARR | £1,000K | £1,085K | £1,173K | £1,000K | | + New ARR | £80K | £85K | £90K | £255K | | + Expansion | £30K | £28K | £35K | £93K | | - Churn | -£20K | -£22K | -£18K | -£60K | | - Contraction | -£5K | -£3K | -£7K | -£15K | | Closing ARR | £1,085K | £1,173K | £1,273K | £1,273K | Forecasting each component: New ARR: Source: Sales pipeline Method: Weight by stage probability Pipeline snapshot: | Deal | Value | Stage | Probability | Weighted | |---|---|---|---|---| | Acme Corp | £50K | Negotiation | 75% | £37.5K | | Beta Inc | £30K | Proposal | 50% | £15K | | Gamma Ltd | £20K | Qualified | 25% | £5K | | Delta Co | £40K | Discovery | 10% | £4K | | Total | £140K | | | £61.5K | Expected new ARR from pipeline: £61.5K Add: Deals not yet in pipeline (historical run-rate): £20K Total forecast new ARR: ~£81.5K Expansion ARR: Method: Historical expansion rate × eligible base Historical data: - Average monthly expansion rate: 2.5% of eligible ARR - Eligible ARR (customers >6 months old): £800K - Expected expansion: £800K × 2.5% = £20K Adjust for known upsells: - Customer X committed to upgrade: £10K - Total expansion forecast: £30K Churn: Method: Historical churn rate × ARR base Historical data: - Average monthly gross churn rate: 1.8% - Current ARR: £1,000K - Expected churn: £1,000K × 1.8% = £18K Adjust for known risks: - Customer Y signalled potential churn: £5K (weight 50% = £2.5K) - Total churn forecast: £20.5K Contraction: Method: Historical contraction rate Historical data: - Average monthly contraction: 0.5% of ARR - Expected: £1,000K × 0.5% = £5K **Top-down forecast model** When to use top-down: - Long-range planning (12-36 months) - Board presentations - Fundraising projections - Scenario planning Method: Apply growth rates to current ARR Simple growth model: Current ARR: £1,000K Historical monthly growth rate: 8% | Month | ARR | Monthly growth | |---|---|---| | Jan | £1,000K | base | | Feb | £1,080K | +8% | | Mar | £1,166K | +8% | | Jun | £1,469K | +8% | | Sep | £1,851K | +8% | | Dec | £2,332K | +8% | Annual growth: 133% (£1M → £2.33M) Adjusted growth model (growth deceleration): Growth typically decelerates as ARR increases: - £0-1M: 10-15% monthly - £1-5M: 5-10% monthly - £5-20M: 3-5% monthly - £20M+: 2-3% monthly Apply deceleration: | Quarter | Starting ARR | Monthly growth | Ending ARR | |---|---|---|---| | Q1 | £1,000K | 8% | £1,260K | | Q2 | £1,260K | 7% | £1,543K | | Q3 | £1,543K | 6% | £1,837K | | Q4 | £1,837K | 5% | £2,127K | Annual growth: 113% (more realistic with deceleration) **Scenario planning** Three-scenario model: Base case (most likely): - Assumptions: Current growth trajectory continues, moderate improvement in retention, planned hiring executed - Monthly growth: 7% declining to 5% - Year-end ARR: £2,127K Upside case (optimistic): - Assumptions: New product launch successful, large enterprise deal closes, expansion rate improves - Monthly growth: 10% declining to 7% - Year-end ARR: £2,800K Downside case (conservative): - Assumptions: Market slowdown, key customer churns, hiring delayed - Monthly growth: 4% declining to 3% - Year-end ARR: £1,500K Scenario probability: - Upside: 20% probability - Base: 60% probability - Downside: 20% probability Expected value: (20% × £2,800K) + (60% × £2,127K) + (20% × £1,500K) = £2,136K Use for: - Budget: Base case - Fundraising deck: Base case with upside potential - Cash planning: Downside case (conservative) - Board reporting: All three with probability weights **Forecast accuracy tracking** Monthly tracking template: | Month | Forecast | Actual | Variance | % error | |---|---|---|---|---| | Jan new ARR | £80K | £75K | -£5K | -6.3% | | Jan expansion | £30K | £35K | +£5K | +16.7% | | Jan churn | £20K | £18K | -£2K | -10.0% | | Jan net new | £85K | £87K | +£2K | +2.4% | MAPE (Mean Absolute Percentage Error): MAPE = Average of |% errors| = (6.3% + 16.7% + 10.0% + 2.4%) / 4 = 8.85% (good — target <15%) Forecast bias analysis: Over-forecasting (optimistic bias): - Sales team consistently overestimates pipeline - Action: Apply haircut to pipeline (e.g., reduce by 15%) Under-forecasting (conservative bias): - Team sandbagging to beat forecast - Action: Adjust stage probabilities upward Track bias by sales rep: | Rep | 3-month avg forecast | 3-month avg actual | Bias | |---|---|---|---| | Rep A | £50K | £42K | -16% (optimist) | | Rep B | £40K | £38K | -5% (accurate) | | Rep C | £35K | £45K | +29% (sandbagger) | Adjust rep-level forecasts by their historical bias **Stage probability calibration** Quarterly calibration: Analyse deals that closed in last quarter: | Stage | Deals entering | Deals closed | Actual probability | |---|---|---|---| | Discovery | 100 | 8 | 8% | | Qualified | 60 | 12 | 20% | | Proposal | 35 | 14 | 40% | | Negotiation | 20 | 15 | 75% | | Verbal commit | 16 | 15 | 94% | Compare to assumed probabilities: - Discovery: Assumed 10%, actual 8% → Adjust down - Qualified: Assumed 25%, actual 20% → Adjust down - Proposal: Assumed 50%, actual 40% → Adjust down - Negotiation: Assumed 75%, actual 75% → Keep - Verbal: Assumed 90%, actual 94% → Keep Impact of calibration: Before calibration: - Pipeline: £500K at Discovery - Weighted: £500K × 10% = £50K After calibration: - Same pipeline: £500K at Discovery - Weighted: £500K × 8% = £40K Difference: £10K per month forecast (significant over time) **Advanced forecasting techniques** Cohort-based forecasting: Instead of single churn rate, use cohort-specific rates: | Cohort age | Monthly churn rate | ARR in cohort | |---|---|---| | 0-3 months | 3.0% | £200K | | 3-6 months | 2.5% | £150K | | 6-12 months | 1.5% | £250K | | 12-24 months | 1.0% | £200K | | 24+ months | 0.5% | £200K | Weighted churn forecast: = (£200K × 3%) + (£150K × 2.5%) + (£250K × 1.5%) + (£200K × 1%) + (£200K × 0.5%) = £6K + £3.75K + £3.75K + £2K + £1K = £16.5K vs simple average: £1,000K × 1.65% = £16.5K (happens to match, but cohort method is more accurate when cohort mix changes) Seasonal adjustment: SaaS often has quarterly patterns: - Q1: Slowest (budget cycle recovery) - Q2: Average - Q3: Below average (summer) - Q4: Strongest (year-end budget flush) Seasonal factors (example): - Q1: 0.85x (15% below average) - Q2: 1.00x - Q3: 0.90x (10% below average) - Q4: 1.25x (25% above average) Apply to forecast: - Average monthly new ARR: £80K - January forecast: £80K × 0.85 = £68K - October forecast: £80K × 1.25 = £100K