Financial scenario model

You help the owner make a decision under uncertainty by showing what the business looks like under different, clearly stated sets of assumptions, and which assumptions actually matter. A good model changes only drivers between scenarios, never outputs; ties out to the cent; and ends in a decision-ready page: "If X happens, cash falls below the buffer in month 8 unless we do Y by month 5."

When to use

• "What if we lose our biggest customer?"
• "Can we afford to hire three more people?"
• "Show me a downside case for the bank."
• "What price increase do we need to break even?"
• "How long is our runway if sales stall?"

What you need

• A starting point: the current budget or the last 6 to 12 months of actuals by month, and the current cash balance (reconciled). Access, best first: a connected app (Xero, QuickBooks, Stripe) on the Connections page; your own browser signed in by the owner to the accounting software; or exports from the owner. If none is there, ask with clarify.
• The decision or question the model must answer, in the owner's words.
• The drivers: revenue base and growth, customer concentration (the top 5 customers' share of revenue), gross margin, fixed costs, planned hires and their cost, debtor days, capital spending, debt repayments.
• The owner's minimum cash buffer and any facility limit.

Method

1. Start from the decision. Write the question at the top of the Summary sheet. Choose the outputs that answer it (lowest cash and when, months of runway, EBITDA, break-even month, covenant headroom).
2. Build the base case from evidence: recent run rate and the signed or recurring business, not hopes. Note the source of each driver.
3. Define scenarios as stories, then numbers. For each scenario write one sentence of what has to be true ("Downside: Cedar, 18% of revenue, leaves in month 4; no growth; customers pay 15 days slower"), then set only the drivers that story changes. Usually: Base, Upside, Downside, plus one or two specific what-ifs the owner asked about. A downside should be one the owner would recognise as plausible and painful, not a catastrophe and not a mild dip.
4. Build one model, many inputs. An Inputs sheet holds every driver in columns per scenario and a Live column picked by a scenario switch (a drop-down cell). The Model sheet reads only the Live column. Never build a separate copy of the model per scenario. Structure and tested code: references/scenario-workbook.md.
5. Model cash, not just profit. At minimum: revenue, gross profit, fixed costs, hires from their start month, EBITDA, working capital (receivables from debtor days), capital spending, debt repayments, closing cash.
6. Compute every scenario in Python too and write the comparison table on a Summary sheet (outputs side by side). Then tie out: for each scenario, set the switch, evaluate the formulas, and compare monthly cash and EBITDA with Python. Any difference is a bug.
7. Sensitivity. Move each driver by a realistic range on the base case (for example plus or minus 10% on price, 20% on volume, 10 days on debtor days) and record the change in the key output. Rank the drivers by impact (a tornado table). Tell the owner which two or three drivers they should watch and manage; that is often more useful than the scenarios themselves.
8. Break-even and runway.
   • Break-even revenue = fixed costs / gross margin %.
   • Months of runway = cash / average monthly net burn (use the forecast burn, not last month's).
   • Required change to hit a target: solve for the driver (for example the price increase that brings EBITDA to zero) with a simple search in Python, and show it.
9. Write the decision page: the question, each scenario in one line with its key outputs, the drivers that matter most, the trigger points to watch (leading indicators with thresholds), and the actions available with their cash effect and timing. The owner decides; you do not.

Worked example: "Can we hire two developers from March?"

Base case from evidence:

• Revenue 50,000 a month, growing 2% a month (the last 12 months' trend).
• Gross margin 60%; fixed costs 26,000 a month; debtor days 35.
• Opening cash 80,000 (reconciled); buffer 20,000; a 15,000 capital purchase in month 2.
• Two hires from month 3 at 4,800 a month each, on-costs included. Scenarios as stories:
• Upside: growth 3.5%, margin 62%, customers pay in 30 days.
• Downside: Cedar (18% of revenue, 9,000 a month) leaves in month 4; no growth; margin 56%; customers pay in 50 days. Results from the model in references/scenario-workbook.md:

  Scenario	Revenue (12 months)	EBITDA (12 months)	Lowest cash	Below the buffer
  Base	670,604	(5,637)	44,887 in month 11	never
  Upside	730,098	44,661	64,270 in month 6	never
  Downside	519,000	(117,360)	(113,200)	month 8
  Sensitivity on the base case: fixed costs plus 10% cost about 33,500 of cash at the low point; growth one point lower costs about 23,800; debtor days plus 10 cost about 3,800. Watch fixed costs and growth; debtor days are not the lever in this business.
  Answer for the owner:

• Base and upside: yes, cash never falls below the buffer.
• Downside: the business runs out of cash in month 8 unless the second hire is paused when Cedar gives notice (saves 4,800 a month) and the facility is drawn before month 7.
• Triggers to watch: notice from any top-5 customer; gross margin below 58% for two months running; fixed costs above 36,000 a month.

Scenario design rules

• Change drivers, never outputs. "Revenue minus 20%" is a result; "Cedar leaves in month 4" is a driver.
• The base case is honest: signed and recurring business only, costs at the current run rate plus committed changes.
• The downside is specific and plausible: one named loss, one slowdown, one slower-paying customer. If the owner says "that could happen", it is the right downside.
• Keep one or two what-ifs that match the decision. Three generic cases plus six what-ifs is a workbook nobody reads.
• Every driver has a source: the last 12 months of actuals, a contract, or "owner assumption" with a date.
• Rerun the sensitivity after any change to the base case; the ranking of drivers can change.

Output

• scenarios-<topic>-<date>.xlsx with sheets: Summary (question, scenario table, decision notes), Inputs (drivers by scenario, switch, sources), Model (monthly, formulas), Sensitivity (tornado table), Checks.
• A show_card with the scenario table: for each scenario, revenue, EBITDA, lowest cash, month cash falls below the buffer (or "never").
• A half-page memo in plain words with the answer and the triggers.

Checks before you finish

• Every scenario differs from base only in its stated drivers.
• For each scenario, the evaluated formulas match the Python model to the cent.
• Opening cash equals the reconciled bank balance.
• Every driver has a source or is labelled as an owner assumption.
• Break-even and runway figures are recomputed from the model, not typed in.
• The sensitivity table uses ranges the owner agrees are realistic.

Pitfalls

• Changing outputs instead of drivers ("downside: profit minus 20%"). Nobody can then say what has to happen.
• Three scenarios that are the same story at different strengths. Make each a distinct, plausible world.
• Forgetting working capital. Growth uses cash: faster growth can lower the cash low point.
• Point estimates presented as forecasts. These are conditional answers: "if this, then that".
• Separate copies per scenario. They drift apart within a week. One model, one switch.
• Probability theatre. Do not attach probabilities unless the owner asks and has a basis for them.

See also

• budget-build, thirteen-week-cash-forecast, unit-economics-model.