Your projection rests on a dozen assumptions: price, volume, costs, growth. None of them are certain. Sensitivity analysis doesn't try to predict which one will fail — it tells you which one matters most if it does. That difference is what separates a professional analysis from an optimistic spreadsheet.
What sensitivity analysis is
An investment project projects revenue, costs, volumes, and growth across several years. Each of those numbers can deviate from the projection. The key question isn't "will it deviate?" — the answer is always yes.
The right question is: if this variable changes by X%, how much does NPV change? That's exactly what sensitivity analysis measures.
It's not about predicting the future. It's about knowing which risks are critical and which are tolerable.
Want to skip the theory and see your own project's tornado? Load the data and you'll get the ranking of critical variables in minutes.
The ±10% method: vary one at a time, hold everything else constant
The standard approach is to shift each variable by ±10% while keeping the others fixed (ceteris paribus). Then observe how much NPV changes. Repeat for every variable and sort by impact.
ΔNPVₓ = NPV(X · 1.10) − NPV(X · 0.90)
Donde
- X
- Variable analyzed (price, volume, cost, etc.)
- ΔNPVₓ
- Range of NPV impact when X varies by ±10%
Desarrollo
- 1.1. Calculate the base NPV with all original parameters.
- 2.2. For each variable X (price, volume, variable cost, fixed cost…):
- 3. a. Raise X by 10% → recalculate NPV₊
- 4. b. Lower X by 10% → recalculate NPV₋
- 5. c. Record ΔNPVₓ = NPV₊ − NPV₋
- 6.3. Sort variables from largest to smallest |ΔNPVₓ| → tornado chart.
Example: a local delivery app
Suppose a project with a base NPV of $85,000. We shift each assumption by ±10%:
| Variable | NPV with −10% | NPV with +10% | Range |
|---|---|---|---|
| Service price | $42,000 | $128,000 | $86,000 |
| Transaction volume | $49,000 | $121,000 | $72,000 |
| Variable cost per transaction | $62,000 | $108,000 | $46,000 |
| Monthly fixed costs | $74,000 | $96,000 | $22,000 |
| Growth rate | $79,000 | $91,000 | $12,000 |
The conclusion is direct: price is risk number one. A 10% drop in price destroys more value than the same move in any other variable. Growth rate, on the other hand, barely moves the needle — even on 5-year projects.
If you can't afford to be wrong about a single variable, that's the one at the top of the tornado. Before investing, validate that variable as if the whole project depended on it. Because in a way, it does.
The tornado chart: the industry's standard image
The tornado sorts variables from highest to lowest sensitivity. The top bars are the widest and represent the highest risk. Visually, the chart narrows from top to bottom — hence the name. Any self-respecting financial presentation includes one.
In Factibilidad.io, the tornado is generated automatically for your project, alongside the table of exact impacts.
Sensitivity vs. scenarios: not the same thing
They're complementary tools, not interchangeable.
- Sensitivity (univariate). Moves one variable at a time. Identifies which one matters most in isolation.
- Scenarios (multivariate). Moves several variables at once. Simulates real situations like an "economic crisis" — where price typically falls, costs rise, and volume drops simultaneously.
Sensitivity analysis answers "which variable is the most dangerous?". Scenarios answer "what happens if everything goes wrong at once?". You need both.
Factibilidad.io includes three preconfigured scenarios — base, optimistic (+10% price, +15% volume), and pessimistic (−15% price, −20% volume, +10% costs) — alongside the tornado.
What to do with the results
Identify critical variables
If price comes out on top of the tornado, your biggest threat is market-related: that the public won't accept your price. The operational conclusion: validate price before investing, not after. Surveys, pilot tests, competitive analysis — whatever it takes.
Set tolerance ranges
Key question: how far can variable X drop
before NPV turns negative?
Example: price can drop by up to 32% before NPV = 0.
If the market is competitive and prices typically fall 15–20%,
the project has enough buffer.
Knowing the exact threshold lets you decide with numbers, not gut feel. "I can sustain up to a 30% price drop" is operational information.
Design contingencies
- If variable cost has high impact: long-term supplier contracts, fixed prices, hedging.
- If volume is critical: validate sales traction before scaling the initial investment.
- If the fragility is in the base break-even, the issue isn't risk — it's structure. Costs need to be redesigned before moving forward.
Why investors ask for this analysis
When you present to an experienced investor, the first thing they do is hunt for the weak point. Sensitivity analysis shows them you found it first — and that you have a plan.
It's the difference between an entrepreneur who says "we're going to sell a lot" and one who says "price is our biggest risk, so we already validated it with 30 pilot customers." The second version closes rounds. The first doesn't pass the first filter.
Apply what you read — load your project and look at your own tornado. The top three variables are the ones you most need to validate before investing.
¿Te fue útil? Compartilo