Finance and the board · Answered by Golden Section from more than 400 B2B software companies observed
Forecast ARR as a bridge, not as a growth rate. Start from ending ARR on a schedule that reconciles to the ledger, then forecast each component from its own driver. Churn and contraction come from the renewal calendar in your contract register and your cohort gross retention. Expansion comes from cohort history. New ARR comes from qualified pipeline by stage times measured conversion, capped by the number of ramped sellers you will actually have. Use worse-than-average assumptions unless your own history proves better, and let rates drift toward benchmarks. Keep the board-approved budget fixed as the plan of record, and run a rolling forecast against it each month, updated with actuals. Headcount and cash then follow from the ARR forecast. Next step: rebuild last year's forecast this way and compare it with what actually happened.
Forecast each ARR component from evidence, never from the target. The budget is the commitment the board measures you against; the rolling forecast is your current best estimate; the gap between them is the conversation, and a gap beyond 10% needs an explanation.
| Metric | Value | What it means | Source |
|---|---|---|---|
| Acceptable budget variance | within 10% | actual result vs board-approved budget, by line, for the yearbeyond it, re-forecast and explain the component that moved | Golden Section operating viewGolden Section operating view |
| Rolling forecast horizon | 18 months | forward period covered by the rolling forecast, updated monthly with actualsrun beside the fixed budget, never in place of it | Golden Section operating viewGolden Section operating view |
| Sales efficiency in a plan | top quartile struggles to beat $0.75 | sales and marketing expense per $1 of new ARR bookeda plan assuming much better than this is likely to miss on revenue and throw every other line out of balance | Golden Section playbookBudget Creation |
| Sales efficiency target | 0.70 or less | prior-period S&M expense ÷ new ARR booked, including upsellsuse your trailing actual, not the target, as the forecast input | Golden Section playbookSales Efficiency Ratio |
| Private B2B SaaS gross revenue retention | 84% median (2025) | starting ARR retained after churn and contraction, excluding expansionBenchmarkit data cited by Golden Section; down from 88% a year earlier at every quartile, so revert churn assumptions toward it | Golden Section, publishedInvesting in Software addendum |
| Forecast sanity check | under $2M to reach $15M is unlikely | total cash a model says the company needs to reach $15M in revenuepossible only with proof of top-10% performance; otherwise rebuild the model | Golden Section playbookCash Flow Forecast |
| Forecast horizon in a data room | 18 months | forward forecast a buyer expects to see beside budget against actuala practical length for the rolling forecast | Golden Section playbookExit Data Room |
A growth-rate forecast hides the parts that move independently. Churn comes from renewals already on the calendar, expansion from how cohorts have behaved, and new ARR from pipeline and selling capacity. Forecast them together and a weak renewal season disappears inside a hopeful bookings number until the quarter is gone. Forecast them apart and you can see which assumption broke, which is the only useful thing a forecast miss can tell you.
The usual error is hope built in at the start. Our budget creation play describes a management team whose most pessimistic budget assumed $0.45 of sales and marketing per dollar of new ARR, when the top quartile struggles to beat $0.75. Every expense line was then sized to revenue that was never coming. Check the plan's implied sales efficiency against your own trailing number before anything else.
The ARR schedule gives the starting point and cohort history. Pipeline management keeps stage data honest by pushing stalled deals back. The cash flow forecast turns the ARR forecast into hiring and cash, updated monthly with actuals, so the model gets more accurate rather than more defended. For the longer horizon, Looking Glass, our venture intelligence platform, forecasts automatically from actual performance, proprietary benchmarks from our portfolio and external benchmarks, which keeps an 18-month view directionally accurate.
All numbers are invented. A company starts the year at $5M ARR and the board wants 40% growth, $2M of net new ARR. The bridge says otherwise. Cohort gross retention of 88% puts about $600K at risk at renewal, and expansion history suggests $300K, so new ARR would need to reach $2.3M. Four ramped account executives at a measured $450K each produce about $1.8M, and two hires starting in March will add little before the third quarter. The forecast lands near $1.5M of net new ARR, about 30% growth. The board approves a 30% budget as the plan of record, funds one more hire on a dated trigger, and receives a rolling forecast each month that shows which component is moving.
Usage-based and AI inference revenue cannot be forecast from contracts; forecast it separately from usage trends and keep it off the committed ARR line. A company with no stage history should write its conversion assumptions down explicitly and expect a wide error until a few quarters of data exist.
From the Golden Section mistakes list, each paired with the play that prevents it.
A forecast that assumes today's growth holds is forecasting a curve that usually fades.
Counting unsigned deals as won inflates the starting point and every month after it.
A model left on last quarter's assumptions stops forecasting and starts defending.
In the order we would run them. Each is on its own page, most with a free Excel template.
Gives the reconciled starting ARR and the cohort history that drives churn and expansion assumptions.
Keeps stage data trustworthy enough to forecast bookings from.
Tests whether the plan's spend can produce the new ARR it assumes.
Turns the ARR forecast into hiring, cash and runway, updated monthly with actuals.
Fixes the realistic plan of record the rolling forecast is compared with.
Executive plays The planning, budgeting and cash plays that a forecast feeds and is checked against.
Accurate enough that the team expects to hit it, which means realistic rather than hopeful. Within 10% is an acceptable variance. Track variance monthly by owner, and treat anything beyond 10% as a reason to re-forecast and explain. A budget missed in the same direction every quarter is miscalibrated, and the fix is how it is built, starting with sales efficiency and retention assumptions anchored to history and benchmarks.
Yes, alongside a fixed budget rather than instead of one. The budget stays as the plan of record the board approved; the rolling forecast, roughly 18 months forward and updated each month with actuals, shows where the year is actually heading. The difference between the two is what the board should discuss.
From qualified pipeline by stage, multiplied by your measured stage-to-close conversion and timed by your measured cycle length. Push stalled deals back a stage before forecasting, count only signed deals as won, and cap the result at what ramped sellers can carry.
Work back from the ARR forecast. Sales hires follow from the new ARR target divided by measured productivity per ramped seller, with ramp time included. Delivery and support hires follow from customer count and implementation load. Everything else should trace to a named strategic initiative in the budget.
Start with unit economics: what a customer costs to acquire, what and when they pay, what they cost to serve, and how they churn or expand. Drive revenue from the ARR bridge, costs from headcount and vendors, and cash from collections timing. Link every assumption to its evidence and give one person ownership of updating it.
Our equity diligence tests a forecast against your history of hitting earlier ones, and our lending is underwritten on the quality of recurring revenue, so a modest forecast you keep beating is worth more in either conversation than an ambitious one you miss. Portfolio companies also get Looking Glass, our venture intelligence platform, which forecasts results automatically from actual performance and from portfolio and external benchmarks.
Talk to Golden Section →Reviewed by Dougal Cameron, CEO & Co-Founder on 2026-10-22. Golden Section observations are labeled separately from external benchmarks and illustrative arithmetic.