SaaS modeling is different from generic startup modeling for one structural reason: revenue is a stock, not a flow. This month's revenue is mostly last month's revenue, adjusted at the margins by new sales, churn, and expansion — which means the model's engine isn't a sales forecast, it's a recurring-revenue waterfall, and the assumptions that matter most are retention assumptions. Get that architecture right and everything a SaaS investor asks about falls out of the model naturally; get it wrong and no amount of formatting rescues it. Here's the structure, the metrics that must be native to it, and how to evaluate whether a SaaS model — yours or one you're diligencing — actually holds up.
The core of every credible SaaS model is one repeating identity: Ending MRR = Beginning MRR + New + Expansion − Contraction − Churn, computed monthly, by cohort or segment where the business has meaningfully different ones. Each component gets its own driver chain — new MRR from the acquisition funnel (pipeline, win rates, or signups and conversion, depending on motion), expansion and churn from retention behavior — and everything downstream references the waterfall. Two modeling implications follow directly. First, small retention assumptions compound violently: 2% versus 3.5% monthly logo churn looks like a rounding difference in the inputs and diverges into entirely different companies by month 36, which is why churn deserves the most scrutinized cell in the workbook. Second, growth in a SaaS model is properly an output — if you find yourself typing a growth rate rather than deriving it from new-minus-churned, the waterfall is decorative.
SaaS models must carry two parallel revenue realities: recognized revenue (earned ratably as service is delivered, per ASC 606) and billings/collections (when cash actually arrives). Annual prepaid contracts drive the two dramatically apart — a $120K annual deal is one month's spectacular cash event and twelve months of $10K recognized revenue, with the difference parked in deferred revenue. A model that conflates them will misstate both profitability and runway, in opposite directions at different times. Structurally: model bookings and billing terms explicitly, derive recognized revenue and the deferred revenue balance from them, and compute burn and runway strictly off the cash line. The prepay-mix assumption doubles as a strategy lever worth scenario-testing — shifting monthly billers to discounted annual prepay is one of the cheapest runway extensions available, and the model should be able to price that trade.
Every SaaS diligence conversation reaches for the same panel, and a well-built model computes them live rather than in a side spreadsheet:
Revenue gets the attention; cost realism is where SaaS models actually fail diligence. Headcount drives everything — model it as a hiring plan with fully loaded costs, and tie sales capacity to the revenue engine (new ARR requires reps, ramped, at quota assumptions you can defend). Scale infrastructure and support costs with customers and usage rather than flat-lining them. And know your market's cost culture: a premium vertical-SaaS product and a race-to-the-bottom horizontal tool imply different tolerable CACs, margin structures, and burn profiles — a model calibrated to the wrong culture produces plans the market won't fund.
Because retention compounds, SaaS scenario analysis is mostly retention analysis. Build the assumptions tab so base/upside/downside cases flex the variables that actually move outcomes — churn, NRR, sales ramp, prepay mix, a delayed raise — and read the downside case for its one critical output: the month cash crosses zero if the pessimistic assumptions hold. That date, tracked as actuals replace forecast each month, is the model's ongoing gift: an early-warning system with enough lead time to act. The craft rules that keep the workbook auditable through all this — assumption cells, color coding, formula consistency — are in financial modeling do's and don'ts.
The diligence checklist we'd run on any SaaS model, useful as a self-audit: Is growth derived from drivers or typed in? Are churn and NRR cohort-based and consistent with the actuals? Does deferred revenue exist and behave correctly against the billing assumptions? Do the metrics recompute when assumptions change, or do they live in a disconnected summary? Is the hiring plan consistent with the sales capacity the revenue requires? And the meta-question — do the assumptions trace to evidence, or to the number the raise needed? A model that passes those questions doesn't just survive diligence; it runs the company between rounds. The generic construction sequence lives in how to build a startup financial model — and building a diligence-grade SaaS model against a live term sheet is precisely the engagement where a SaaS-fluent fractional CFO earns the fee.