← Back to articles

Sensitivity Analysis: Pre-Building the Questions Investors Will Ask



What Is Sensitivity Analysis in a Startup Financial Model?

Sensitivity analysis shows how a change in one or more input assumptions flows through to a change in the model's key outputs --- typically runway, gross margin, and revenue. It is distinct from scenario planning (which changes multiple assumptions together to reflect different business environments) and is instead focused on isolating the impact of individual variables.

Key distinctions:

What	One variable at a time	Multiple variables together changes
Purpose	Identify which inputs drive the Show coherent future states most variance
Output	Tornado chart or sensitivity	Conservative / Base / table	Aggressive cases
Best for	Identifying risk concentration Investor-facing narrative Builds	Model robustness	Investor confidence

Why Investors Use Sensitivity Analysis in Diligence

When a VC analyst goes through a financial model in diligence, one of the standard checks is to change the key assumptions by 10-20% and see what happens to the bottom line. This is not adversarial. It is how experienced investors understand which assumptions the model is most dependent on.

If the model has not been built with this flexibility in mind --- if assumptions are hard-coded rather than centralised --- the analyst cannot run the test and flags this as a model quality issue. If the model has sensitivity analysis pre-built, the analyst can see the founder has already done the work and understands their own key risks. The question "what are the two or three assumptions your model is most sensitive to?" is one of the most common investor diligence questions. Pre-building the sensitivity is the equivalent of pre-building the answer.

The Four Most Important Sensitivities for Early-Stage Models

1. CAC (Customer Acquisition Cost)

If CAC is 20% higher than modelled, what happens to marketing spend, to the CAC payback period, to LTV:CAC, and to monthly burn? For most early-stage businesses, CAC is one of the most uncertain and highest-impact variables. Build a sensitivity that shows the impact across a range of CAC scenarios.

2. Churn rate

If monthly churn is 1 percentage point higher than modelled, what happens to net revenue retention, to total customer count at year-end, and to the revenue line? For SaaS businesses especially, the compounding effect of churn means a small change in churn rate produces large changes in the revenue line over 18-24 months.

3. Revenue growth rate

If the core growth assumption is 20% below plan, what is the revised runway? What is the new point at which the company needs its next round? This is often the most powerful single sensitivity because it directly addresses the question investors most want answered: what is the downside scenario?

4. Gross margin

If COGS are 10% higher than modelled (due to higher infrastructure costs, pricing pressure, or operational inefficiency), what happens to gross margin and operating leverage? For businesses targeting high gross margins, the sensitivity shows how robust the unit economics are to cost pressure.

How to Build a Sensitivity Table in Excel

A sensitivity table (or data table in Excel terminology) shows the output of a model formula across a range of input values. The structure is straightforward:

One-variable sensitivity (1D data table):

Column header: range of values for the input (e.g. CAC from $200 to

$400 in $25 increments)

Row output: the metric being measured (e.g. 18-month runway, gross margin)

Built using Excel's Data > What-If Analysis > Data Table Two-variable sensitivity (2D data table):

Row header: range of values for input A (e.g. CAC)

Column header: range of values for input B (e.g. churn rate) Cell values: output metric (e.g. monthly burn at month 18) Built the same way but using both row input cell and column input cell

The key requirement for both: the input must flow from a single cell reference in the model (not be hard-coded in multiple places). Centralised assumptions tabs are what make sensitivity analysis possible without rebuilding.

Key insight: Sensitivity analysis is only possible if the model is built with centralised assumptions. A model where growth rates, CAC, and churn are hard-coded in multiple tabs cannot be stress-tested efficiently. The assumptions tab is what makes sensitivity analysis fast.

The Tornado Chart: Visualising Which Inputs Matter Most

A tornado chart ranks input variables by the size of their impact on a key output (typically revenue, runway, or gross margin). The widest bar is the variable the business is most sensitive to. The narrowest is the least critical.

Building a tornado chart requires running a high and low test for each variable (e.g. ±20%) and recording the change in output. The result is a ranked list that immediately communicates the model's key risk concentrations.

For most early-stage companies, the ranking looks roughly like this: 1. Revenue growth rate (widest bar, highest impact)

2. Churn rate (for subscription businesses)

3. CAC (for sales-led or performance-marketing-driven businesses) 4. Gross margin

5. Headcount growth (typically lower impact in early models) A tornado chart in the investor deck communicates two things simultaneously: the founder has done the work of understanding their key uncertainties, and the business is not dependent on a single fragile assumption.

Common Sensitivity Analysis Mistakes

Running sensitivities on outputs that are already robust, ignoring the fragile ones

Hard-coding assumptions that prevent efficient testing

Showing only upside sensitivities (what happens if CAC is lower than expected)

Not including sensitivity results in investor materials, doing the analysis only internally

Sensitivity ranges that are too narrow to reflect realistic uncertainty

Summary

Sensitivity analysis turns a financial model from a static forecast into a tool for understanding risk. The four variables most worth testing in early-stage models are CAC, churn, revenue growth rate, and gross margin. Build sensitivity tables using centralised assumption cells, visualise the results in a tornado chart for investor materials, and use the analysis to pre-answer the questions investors will ask. A founder who has run their own sensitivities and can speak to the results is in a structurally stronger position in any diligence conversation than one who has not.

Common Mistakes Founders Make During Fundraising

The most expensive fundraising mistake is starting too late. Most founders begin outreach when they have 3-4 months of runway, which means they are negotiating from a position of desperation rather than strength. The rule of thumb: start fundraising when you have 9-12 months of runway, which gives you time to be selective, build relationships before asking, and walk away from bad terms.

The second most common mistake is treating all investors as interchangeable. A $1M cheque from a generalist angel who does not understand your space is materially less valuable than the same cheque from a domain-expert who can open doors, advise on hiring, and provide credibility with the next round's investors. Spend time mapping which investors have backed comparable companies and who can genuinely add value beyond capital.

Sharing your financial model too early before you understand what narrative it supports is another frequent error. Investors will poke at your assumptions; if you have not stress-tested your own model, you will be caught flat-footed. Run your own sensitivity analysis before sharing. Know which assumptions drive the outcome, which are defensible, and which are genuinely uncertain and why you have chosen your specific estimate.

Finally, many founders fail to maintain competitive tension. Investors move faster when they know others are interested. Running a tight, parallel process meeting multiple investors in the same 4-6 week window is not rude; it is expected professional behaviour. Telling an investor you have other conversations at a similar stage is appropriate; it signals that the opportunity is competitive.

What Investors Are Actually Evaluating

Early-stage investors particularly pre-seed and seed are making a bet on the team before there is sufficient evidence to bet on the business. The three questions they are answering are: can this team build what they say they are building, can they sell it, and can they raise again? Everything in your pitch, your data room, and your financial model feeds these three questions.

At Series A, the emphasis shifts toward evidence of product-market fit and the beginnings of repeatable unit economics. Investors at this stage want to see cohort data showing retention, CAC by channel broken out from blended numbers, NRR above 100% for SaaS, and a clear model for how spending $X in sales and marketing generates $Y in predictable ARR.

Soft signals matter too. Responsiveness, clear communication, and handling difficult questions well all feed into an investor's assessment of whether they want to work with this team for the next 7-10 years. Founders who over-explain, become defensive about their model, or cannot answer basic questions about their own business quickly undermine confidence.

The Most Common Financial Modeling Mistakes

The most dangerous mistake in startup financial modeling is building a model that only works in one scenario. Real businesses face unexpected churn, slower-than-expected sales cycles, competitive pricing pressure, and hiring delays. A model that only shows the plan without stress testing what happens if ARR growth is 30% lower, or if a key hire takes four months to land is not a planning tool; it is a wishful thinking exercise.

Circular references are a technical trap that undermine model credibility instantly. When an investor opens your spreadsheet and sees #REF errors or formula loops, it signals that the model has not been rigorously tested. Build revenue, cost, and cash flow on separate sheets with clear linking. Every input assumption should live in a dedicated assumptions tab so an investor can change your growth rate and see the full impact cascade through the model instantly.

Overcomplicated models are as problematic as oversimplified ones. A 40-tab model that takes 20 minutes to navigate tells an investor that the builder does not understand what drives their business. The best financial models are opinionated: they make clear which 3-5 assumptions matter most, and they are built to make sensitivity analysis on those assumptions easy.