Short answerSensitivity analysis changes one variable at a time to show which driver moves your model most; scenario analysis changes several variables together to describe a coherent future — base, best and worst case. Run sensitivity first to find your five to seven highest-impact drivers, then build scenarios around those drivers.
Most founders show investors one forecast and call it "conservative." Investors don't believe it — and more importantly, they notice what's missing. Standard Ledger's April 2026 review of investor red flags puts "no scenario planning" among the five model flaws that kill confidence fastest. This piece gives you a working scenario analysis financial model framework: what sensitivity and scenario analysis each do, why the order matters, and how to build three cases you can defend in a partner meeting without flinching.
What is the difference between scenario and sensitivity analysis?
The two techniques answer different questions, and mixing them up produces analysis that answers neither.
Sensitivity analysis flexes one input while holding everything else constant. What happens to runway if churn rises from 3% to 4%? What happens to gross margin if hosting costs climb 20%? It is a microscope: it isolates each assumption and measures its individual pull on the output. Finance Alliance's March 2026 comparison describes the standard workflow — set a base case, pick key inputs, vary each by ±10–20%, and read the results off a sensitivity table or tornado chart.
Scenario analysis flexes several inputs at once to describe a plausible future. In a recession case, new bookings slow and churn rises and your next round takes longer — because in reality, bad things travel in packs. A scenario is a story told in numbers, not a dial turned in isolation.
| Sensitivity analysis | Scenario analysis | |
|---|---|---|
| Variables changed | One at a time | Several, simultaneously |
| Question answered | Which driver matters most? | What would this future do to us? |
| Typical output | Tornado chart, data table | Base / best / worst P&L and cash view |
| Best for | Ranking risks, validating the model | Strategic decisions, stress tests, fundraising |
| Failure mode | Ignores how variables interact | Cases picked arbitrarily, no logic behind them |
Neither replaces the other: sensitivity ranks the levers, scenarios show what a full future does to your cash.
Why you should run sensitivity analysis first
Here's the sequencing most founders skip: sensitivity analysis is the filter that decides what your scenarios are made of.
A typical startup model has 40–80 assumptions. You cannot build scenarios around all of them, and you shouldn't try. G Squared Partners' December 2025 guidance on SaaS forecasting is blunt about over-engineering: models with dozens of flexed variables become unmaintainable, so focus on the five to seven drivers that create roughly 80% of the financial impact. Finance Alliance reaches the same conclusion from the other direction — start with sensitivity to identify critical variables, then build scenarios around combinations of those drivers, so you never waste effort modeling futures that hinge on numbers that barely move the output.
For an early-stage software company, the shortlist usually looks like this:
- New customer additions (or pipeline conversion rate)
- Churn / net revenue retention
- Average revenue per account and pricing
- Sales and marketing spend (and its efficiency)
- Hiring pace and fully loaded payroll
- Payment timing — when cash actually lands versus when revenue books
Flex each across a realistic range. Whatever swings your cash-out date by more than a month earns a place in your scenarios. Everything else stays fixed at base-case values.
How to build base, best and worst cases you can defend
"Defensible" has a precise meaning here: every number traces to a source, every case tells one coherent story, and every case ends in a decision. Five steps.
1. Anchor the base case in trailing actuals
Your base case is not your pitch-deck plan. It is what happens if current performance simply continues. G Squared Partners recommends trailing three-to-six-month averages for churn and expansion precisely because they smooth out one-off spikes. If your last two quarters say 4% monthly churn, your base case says 4%.
2. Change only the shortlisted drivers
All three cases should flex the same five to seven drivers and nothing else. This is what makes cases comparable: an investor can lay them side by side and see exactly which assumptions separate the futures. When each scenario quietly rewrites different parts of the model, comparison becomes impossible and trust evaporates.
3. Write the story before the numbers
Each case needs a narrative you can say out loud. Best case: "The new onboarding flow cuts early churn by a fifth and the Q3 pricing change lands." Worst case: "The enterprise segment freezes budgets and our Series A slips six months." If you can't name the mechanism, you're just scaling revenue by ±20% — which is decoration, not analysis.
4. Attach a trigger and an action to every case
This is the step that separates a scenario model from a slide. For each case, define the leading indicator that tells you it's materialising and the action you've pre-committed to take. G Squared Partners frames the downside case exactly this way: it exists to set your minimum cash requirement and trip predetermined responses — cut burn, shift spend from acquisition to retention, reprice.
5. Pressure-test the worst case against runway
Run the downside through your cash flow, not just your P&L. If the worst case ends with zero cash before your next realistic fundraise window, you don't have a scenario problem — you have a runway problem, and you've just discovered it while it's still cheap to fix. Pair this exercise with a 13-week cash flow forecast for the near-term view.
A worked example
A B2B SaaS at €150K MRR, growing €12K net new per month, with 9 months of cash:
| Driver | Worst | Base | Best |
|---|---|---|---|
| New MRR / month | €7K | €12K | €16K |
| Monthly gross churn | 4.5% | 3.5% | 2.8% |
| Hiring in next 2 quarters | Freeze | 4 roles | 6 roles |
| Round closes | +6 months | On plan | On plan, larger |
| Cash-out date | Month 8 | Month 13 | Month 16+ |
| Trigger | 2 straight months of churn >4% | — | Onboarding cohort churn −20% |
| Pre-committed action | Freeze hiring, cut paid spend 30% | Review monthly | Pull 2 hires forward |
The founder who walks into a partner meeting with this table isn't predicting the future. They're demonstrating they'll notice which future is arriving — and already know what they'll do about it.
The mistake that makes scenarios useless
Here's the decision rule we'd put on a sticky note: a scenario without a trigger is a slide, not a plan.
Most scenario decks fail not because the numbers are wrong but because nothing is wired to them. The worst case lives in the appendix; six months later churn has been above threshold for a quarter and nobody noticed, because no one defined the threshold. G Squared Partners' early-warning framing makes the fix concrete: if monthly churn exceeds baseline by more than one percentage point for two consecutive months, that's not a data point — that's your downside case knocking. The same logic powers a proper downturn scenario plan: scenarios earn their keep on the day reality picks one.
There's also a signalling effect. Standard Ledger notes that a genuine downside case doesn't make you look pessimistic to investors — it makes you look like someone managing a real business with real risks. A model that only shows the bull case reads as a model that was never tested. It's one of the classic financial-model mistakes that make investors close the tab.
What this means for you
Sequence the work: sensitivity first, scenarios second, triggers always. Flex every assumption once, keep the five to seven that move your cash-out date, and build exactly three cases on top of them — trailing actuals for the base, named initiatives for the upside, correlated bad news for the downside. Then wire each case to a threshold you check monthly and an action you've already agreed with your team. That's the difference between a forecast investors skim and the financial model VCs actually read.
Frequently asked questions
How many scenarios should a startup financial model include?
Three is the working standard: a base case built on trailing actuals, an upside case tied to named initiatives, and a downside case that stress-tests runway. More than four and nobody maintains them. Each scenario should change the same handful of drivers, so investors can compare cases line by line.
What should a worst-case scenario assume?
Realistic bad news, not catastrophe: slower customer acquisition, higher churn, longer sales cycles, a fundraise that slips two quarters. Standard Ledger's April 2026 guidance notes a downside case signals you manage real risks — investors read its absence as a model that was never stress-tested.
How often should founders update their scenarios?
Refresh the base case monthly when actuals land, and rebuild all three cases quarterly or after any material event — a funding round, a large customer win or loss, a pricing change. Stale scenarios are worse than none, because they anchor decisions to assumptions the business has already outgrown.
What is a tornado chart in sensitivity analysis?
A tornado chart ranks your model's variables by how much each one swings the output when flexed across a realistic range, widest bar on top. It turns a spreadsheet full of assumptions into a one-glance answer to "which numbers actually matter?" — and tells you where to focus scenario work and data collection.
Sources
- Finance Alliance — Sensitivity analysis vs scenario analysis, 24 March 2026. https://www.financealliance.io/sensitivity-analysis-vs-scenario-analysis/
- G Squared Partners — SaaS Forecasting Models: Building Scenarios Around Churn and Expansion Revenue, 2 December 2025. https://www.gsquaredcfo.com/blog/saas-forecasting-models
- Standard Ledger — The 5 Red Flags in Your Financial Model That Will Kill Investor Confidence, 27 April 2026. https://www.standardledger.co/uk/article/the-5-red-flags-in-your-financial-model-that-will-kill-investor-confidence
