What is sales forecasting? Predicting revenue when part of it is consumption
Written by Arnon Shimoni
✓ Expert
Last updated on:
Sales prediction analysis, more commonly called sales forecasting, is the practice of estimating future revenue from pipeline data, historical patterns and rep judgement. The traditional version answers one question: how much will we close this quarter.
That question is no longer sufficient. On a consumption-priced product, a large share of next quarter's revenue comes from customers who already signed, and it depends on how much they use rather than on what anyone closes. Forecasting has become two problems that most teams still run as one.
Field | Detail |
|---|---|
What it is | Estimating future revenue from pipeline, history and judgement |
Traditional inputs | Deal stage, close date, deal size, rep confidence, historical win rates |
Additional inputs for consumption | Usage trend by cohort, commit drawdown rate, overage patterns, seasonality |
Two distinct forecasts | New bookings, and consumption revenue from the existing base |
Common failure | Forecasting bookings accurately while consumption revenue swings unforecast |
Related |
The classic approaches
Stage-weighted pipeline. Each stage carries a probability, multiply and sum. Simple, and systematically optimistic because stage is a poor proxy for likelihood.
Historical conversion. Apply observed stage-to-close rates by segment. Better, and needs enough volume per segment to be stable.
Rep commit. Ask reps what will close. Accurate in disciplined teams, and prone to sandbagging and hero-calling in others.
Cohort or velocity models. Forecast from deal flow rate and cycle time rather than from individual deals. Robust at volume, weak on lumpy enterprise pipelines.
Statistical and ML models. Fit on historical deal attributes. Useful once you have thousands of closed deals, misleading before that.
Why consumption pricing breaks bookings forecasting
A bookings forecast predicts contract value at signature. On a subscription that is nearly the same as predicting revenue. On a consumption model the two can diverge sharply in both directions.
A customer signs a modest commitment and consumes three times it. Another signs a large commitment and undershoots it all year. A third grows steadily for two quarters and then migrates a workload away in a week. None of that is visible in a pipeline report, because it happens after the deal is closed and the opportunity is marked won.
The practical consequence is that finance receives a forecast that is accurate about the wrong number. Bookings come in as predicted, revenue does not, and nobody can explain the gap from the CRM.
Forecasting consumption revenue
Consumption revenue is more forecastable than most teams assume, because it is dominated by existing accounts with observable behaviour rather than by uncertain future events.
Separate the two forecasts. New bookings and base consumption are different problems with different inputs. Combining them hides both.
Start from contracted floors. Minimum commits and volume commitments give a revenue floor that is close to certain. See revenue backlog.
Model drawdown rate. How fast each cohort is consuming against its commitment tells you who will exceed, who will undershoot, and roughly when.
Forecast overage separately. Usage above commitment behaves differently from usage within it and is far more volatile.
Use cohort trends, not averages. A blended growth rate across accounts of different ages and use cases predicts nothing.
Track leading indicators. Usage per active account, meters newly triggered, and workload concentration move before revenue does.
What makes a forecast credible
Accuracy matters less than decomposition. A forecast that is ten percent off but explains where the variance came from is more useful than one that lands exactly and cannot be interrogated.
Component | Certainty | Source |
|---|---|---|
Contracted minimums | High | Signed commitments not yet consumed |
Base consumption within commit | Medium-high | Cohort drawdown trend |
Overage above commit | Medium | Historical overage rate by cohort |
Renewals and expansion | Medium | Utilisation and product engagement |
New bookings | Low-medium | Pipeline and historical conversion |
Churn and workload migration | Low | Usage decline signals, support and account signals |
Presenting a forecast in these bands lets the business act on it. A shortfall concentrated in overage is a very different problem from a shortfall in new bookings, and the responses have nothing in common.
Where Solvimon fits
Solvimon meters usage and holds commitments in the same system, so drawdown against contracted minimums is directly observable rather than reconstructed at month end. That makes the contracted floor and the consumption trend available as forecast inputs while the period is still open.
Because charges are calculated continuously rather than at close, a mid-period view of where each account is tracking against its commitment is a report rather than an exercise.
Frequently Asked Questions
What is sales prediction analysis?
It is sales forecasting: estimating future revenue using pipeline data, historical conversion patterns and rep judgement. For consumption businesses it also has to model revenue from usage by existing customers.
Why does usage-based pricing make forecasting harder?
A bookings forecast predicts contract value at signature, but consumption revenue depends on what customers actually use afterwards. Accounts can overshoot or undershoot their commitments substantially, and none of that appears in a pipeline report.
How do you forecast usage-based revenue?
Forecast it separately from new bookings. Start from contracted minimums as a floor, model drawdown rate by cohort to project consumption within commitments, forecast overage separately because it is more volatile, and use cohort trends rather than blended averages.
What is the most accurate sales forecasting method?
There is no single answer, but forecasts decomposed by certainty band outperform single-number forecasts in usefulness. Contracted minimums are near-certain, overage is volatile, and new bookings are the least certain component.
What leading indicators predict consumption revenue?
Usage per active account, the rate at which accounts trigger new meters, drawdown pace against commitments, and workload concentration. These move before revenue does.
Should sales own the consumption forecast?
Usually not alone. New bookings is a sales forecast. Base consumption is closer to a finance and product analytics problem, since it depends on customer behaviour rather than on deals in flight.
Related
Revenue backlog. Contracted revenue not yet recognized.
Minimum commit. The contracted floor a forecast can rely on.
Net revenue retention. How the existing base grows or shrinks.
Sales optimization. Improving what the forecast is measuring.
MRR and ARR. Recurring revenue metrics and their limits for usage models.
Ready for billing v2?
Solvimon is monetization infrastructure for companies that have outgrown billing v1. One system, entire lifecycle, built by the team that did this at Adyen.
