financial-modeler

Overview

Financial models translate business assumptions into numbers: how revenue grows, what it costs, when cash runs out, and what the business is worth. This skill covers building clean, auditable models — revenue build-ups, cost structures, cash flow, scenario analysis — whether in spreadsheets or code.

A model is a thinking tool first, a prediction second. Its value is in exposing which assumptions actually matter.

When to use

• Forecasting revenue for a business plan or fundraise

• Modeling unit economics (CAC, LTV, payback)

• Running scenario analysis (best/base/worst cases)

• Building a hiring or expansion plan tied to revenue

• Valuing a business (DCF, comparables, multiples)

• Stress-testing whether the business works at scale

• Modeling M&A scenarios and synergies

• Building board-ready operating plans

• Stress-testing runway under downside cases

Core concepts

Drivers, not plugs. Every line item should derive from an operational driver (users × conversion × price), not a hardcoded growth percentage. Driver-based models are explainable and testable.

Unit economics. Contribution margin per customer, CAC payback period, LTV:CAC ratio. If unit economics don't work, scale makes things worse. Model these before anything else.

Three statements. P&L (profitability), cash flow (survival — profit ≠ cash), balance sheet (position). Early-stage models can be P&L + cash flow; add the balance sheet as complexity grows.

Scenarios and sensitivity. Base, upside, downside cases. Sensitivity analysis (tornado charts or data tables) shows which assumptions swing the outcome most — focus diligence there.

Model hygiene. One input tab (all assumptions, blue font, sourced), separate calculation tabs, no hardcoded numbers in formulas, consistent time periods, error checks (balance sheet balances, subtotals foot), and version control.

Valuation basics. DCF (project cash flows, discount at WACC, terminal value), comparable companies (revenue/EBITDA multiples), precedent transactions. Triangulate — no single method is truth.

Cohort-based revenue. For subscription or repeat-purchase businesses, model revenue by cohort: new customers acquired per period × retention curve × ARPU expansion. Cohort models capture the compounding truth that simple growth-rate models miss — and they expose when growth is just churning through new logos.

Hiring plan linkage. Headcount is usually the largest cost and the hardest to cut. Tie every hire to a revenue milestone or capacity constraint, model fully-loaded costs (salary × 1.25–1.4 for benefits, taxes, equipment), and show the hiring curve against the revenue curve — investors check that the team scales with, not ahead of, the business.

Driver-based modeling. Every line item traces to operational drivers: revenue = customers × ARPU; support cost = tickets × cost per ticket. Driver-based models answer "what if" questions; plug-number models just describe. Document each driver's source (historical data, benchmark, assumption) and confidence level. Scenario framework. Base (most likely), upside (things go well), downside (things go poorly) — with explicit trigger points for switching plans. The value is not prediction accuracy but preparedness: pre-decided actions for each scenario. Update scenarios quarterly; stale scenarios provide false comfort. SaaS metrics. ARR, NRR (net revenue retention — the single best SaaS health metric), CAC payback, rule of 40 (growth rate + profit margin ≥ 40%), burn multiple (net burn / net new ARR). Know benchmarks for your stage; investors will compare you against them regardless.

Practical workflow

1. Define the purpose. Fundraising model? Operating plan? Valuation? Purpose determines horizon, granularity, and which scenarios matter.
2. Map the revenue engine. Start from the top of the funnel: leads → conversion → customers → ARPU → revenue. Add expansion, churn, and pricing changes for subscription models.
3. Build the cost structure. Fixed vs. variable costs. Headcount plan (the biggest cost for most startups) tied to revenue milestones. COGS tied to revenue drivers.
4. Connect the statements. P&L flows to cash flow (working capital changes, capex) and balance sheet. Verify: cash reconciles, balance sheet balances.
5. Add scenarios. Build a scenario switch (base/upside/downside) driven by 3–5 key assumptions. Run sensitivity on the top drivers.
6. Document and review. Every assumption gets a source or rationale. Add a "key assumptions" summary tab. Have someone else audit the formulas — model errors are embarrassingly common.

Audit checklist: no hardcoded values in formulas, all assumptions on the input tab, balance sheet balances, cash flow ties to cash balance, subtotals cross-foot, scenarios switch cleanly, charts match underlying data.

Unit economics one-pager (build first, before the full model): CAC by channel, gross margin %, LTV (ARPU × gross margin × average lifetime in months), LTV:CAC ratio, CAC payback months, and contribution margin per order/customer. If payback exceeds 12 months for SMB or 18 for enterprise without a clear reason, fix the business before modeling its future.

Model audit checklist: formulas consistent across periods → no hardcoded numbers in formula cells → balance sheet balances → cash flow ties to P&L and balance sheet → circular references resolved intentionally → sensitivity tested on top 5 drivers → version controlled with change log. A model nobody audits is a model nobody should trust. Board package structure: KPI dashboard (one page) → P&L vs. plan with variance commentary → cash and runway → key risks and mitigations → decisions needed. Boards want insight and decisions, not spreadsheets — the model supports the narrative, not the reverse.

Common pitfalls

• Garbage assumptions, precise output. Five-decimal precision on made-up inputs. Be honest about uncertainty; use ranges.
• Hardcoded numbers in formulas. Makes the model brittle and unauditable. All inputs live on the input sheet.
• Ignoring cash flow. Profitable on paper, bankrupt in reality. Model cash timing (receivables, payables, upfront costs).
• Hockey sticks without engines. Growth curves need drivers (proven CAC, expansion revenue). Otherwise they're drawings.
• No scenarios. A single-point forecast is a lie told with confidence. Always show the range.
• Overcomplexity. 40 tabs nobody understands. Model at the granularity decisions require — no finer.
• Forgetting working capital. Growth consumes cash (inventory, receivables). Fast-growing businesses die from this most.
• Mixing cash and accrual. Recognizing annual prepayments as immediate revenue inflates early months. Model cash timing separately from revenue recognition.
• Forgetting the balance sheet. Debt, deferred revenue, and working capital all affect cash. Even a simple balance sheet catches errors the P&L hides.
• False precision. 5-decimal forecasts of unknowable futures. Round aggressively; precision implies certainty you do not have.
• Static models. Building once, never updating. A model not updated monthly is decoration — assign ownership and a cadence.