Building a startup financial model

A startup financial model is not a prophecy. It is a machine that turns explicit assumptions into P&L, cash and runway so you can see what must be true for the plan to work. You build it to decide — hire or not, raise or cut, price-change impact — and to show investors you understand your own economics.

Keep it simple enough that you still believe it six months later. A clear Google Sheet beats a 40-tab cathedral. Always output cash, not just accrual profit. Startups go bankrupt on cash.

Concepts in this section

Read them in order.

Step	Concept	What it answers
1	Driver-based models	What is a driver vs a plugged revenue cell?
2	Revenue build and cohorts	How does funnel × ARPA become ending ARR?
3	Hiring plan in the model	What does a hire actually do to burn?
4	Scenarios and sensitivity	Which knob moves cash the most?

A sales-led month with 200 leads, 25% to opportunity, 20% to close and $12k ARPA is 10 customers and $120k new ARR. On an $800k book with 1.5% monthly churn and 1.0% expansion, ending ARR is $916k. If two AEs can only close ~4 deals/month each, a model that "closes" 10 is lying — add a capacity driver.

People are most of burn. Three engineers at $180k base and a 1.30 load factor cost $58,500/month. On $2.5M cash and $180k prior burn, that cohort alone can cut runway from ~14 months to ~10.5. Run the hire through the downside case before you open the req.

Tie COGS to revenue or usage. Early AI features often pull margin from 80% toward 50–60%. Model it explicitly. Reconcile to bank and billing monthly. If the sheet needs LTV/CAC > 3× to clear cash in 24 months but trailing actuals imply 1.5×, the plan is fiction.

Owners: founder/CEO and a finance-minded operator. Engineers contribute capacity plans, infra COGS and metering. Interview angle: "Walk me through how you'd forecast ARR and cash for the next year." Start from drivers, not from a hockey-stick CSV.

Common mistakes

Plugged revenue you cannot debug.
Hiring in month 1 while revenue ramps in month 12.
100% quota attainment forever. Real teams average 60–80% early.
One scenario only, then surprise when the world is the downside case.
Ignoring payment timing so ARR growth overstates runway.

Key takeaways

Prefer driver-based models over plugged outputs.
Build revenue from funnel × ARPA and evolve ARR with cohort churn/expansion.
Headcount fully loaded cost is the primary burn driver.
Maintain base / upside / downside and sensitivity on a few key drivers.
Reconcile to cash and bank actuals; use the model to gate hires and fundraising timing.