Which assumption breaks your project first? How to build best/base/worst case worlds, measure risk with real numbers, and know exactly how deep a CTO needs to go — no deeper.
Every financial projection rests on assumptions — never a guarantee. Forecasting quality depends on the analyst's experience, market understanding, domain knowledge, operational insight, and vendor/industry estimates. Two analysts can produce two different estimates for the exact same project, simply because their assumptions differ.
Sensitivity and scenario analysis exist precisely because inputs are uncertain, actual outcomes may differ, and management needs to understand the risk before it becomes a surprise.
┌───────────────────────────────────────────────────────────┐ │ PURPOSE: see how sensitive NPV is to ONE input at a time │ │ Question: "Which variable, if wrong, hurts my project most?" │ │ │ │ TYPICAL INPUT VARIABLES: │ │ Revenue benefits · Operating costs · │ │ Initial investment (CAPEX) · Discount rate │ │ │ │ METHOD: │ │ 1. Start with BASE CASE │ │ (e.g. Revenue = 2.5 Cr → NPV = 1.13 Cr, rate = 12%) │ │ 2. Pick ONE input (say revenue), change it ±10% │ │ Revenue −10% → 2.25 Cr | Revenue +10% → 2.75 Cr │ │ 3. Recalculate NPV. Note how much it moves. │ │ 4. Repeat for discount rate, cost, CAPEX, etc. │ └───────────────────────────────────────────────────────────┘
Revenue 2.5 Cr → 2.25 Cr → NPV dropped SIGNIFICANTLY Discount rate 10% → 14% → NPV moved, but LESS CONCLUSION: this project is more sensitive to REVENUE assumptions than to the DISCOUNT RATE.
┌───────────────────────────────────────────────────────────┐ │ SENSITIVITY SCENARIO │ │ ───────────────── ───────────────── │ │ ONE variable changes at a time MULTIPLE variables change │ │ TOGETHER │ │ Objective: find which variable Objective: see what happens │ │ matters most in overall "worlds" │ │ (good / bad / normal) │ └───────────────────────────────────────────────────────────┘
BASE CASE — the most likely view OPTIMISTIC (Best) — high revenue, lower cost PESSIMISTIC (Worst) — low revenue, higher cost, delays (You can add more: Good, Average, Poor, Extreme Worst, etc.)
Expected NPV = Σ (NPV of scenario × Probability of scenario) Example from class: Base: 1.13 × 0.50 Optimistic: 3.04 × 0.25 Pessimistic: −0.15 × 0.25 ──────────────────────────── = POSITIVE overall → project remains viable ON AVERAGE
Stress Testing is the extreme worst-case version — everything goes bad at once. Common in banking (per RBI guidelines), asking simply: "can we survive this?"
Measures how much NPV varies around its EXPECTED value. Higher σ → outcomes more spread out → MORE RISK. Concept (not formula-heavy): 1. Compute expected NPV 2. For each scenario: deviation = (Scenario NPV − Expected NPV) 3. Square it, multiply by that scenario's probability 4. Sum these = VARIANCE 5. Square root of variance = STANDARD DEVIATION Excel: STDEV() works for equal-probability data; probability- weighted scenarios need a manual/custom setup.
CV = Standard Deviation / Expected NPV Used to COMPARE riskiness across different projects. Lower CV = less risk per unit of return. Example: Project A: CV = 0.5 Project B: CV = 1.2 👉 Project A is less risky relative to its return.
Do both when the project is big, strategic, or high-risk — you want both "what matters most" and "what happens if worlds go bad or great."
PRACTICAL GUIDELINE: 1. START WITH SENSITIVITY Identify key drivers: revenue? cost? FX? discount rate? → Decide what to monitor closely. 2. THEN SCENARIO Combine BAD values of key drivers into a pessimistic scenario. Combine FAVORABLE values into an optimistic scenario. Compute expected NPV + risk metrics. 3. ADVANCED (optional, usually finance team's job) Monte Carlo simulations — generate random scenarios at scale using Excel plugins or dedicated tools.
Goal Seek answers: "What's the minimum revenue needed so NPV = 0?"
EXCEL STEPS: 1. Build your normal NPV model 2. Data → What-If Analysis → Goal Seek 3. Set cell = NPV cell 4. To value = 0 5. By changing cell = Revenue (or whichever input) CLASS EXAMPLE: Break-even revenue benefit ≈ ₹2.05 Cr Revenue ≥ 2.05 Cr → project stays viable Revenue < 2.05 Cr → NPV goes negative
TYPICAL FORECAST HORIZONS
1–2 quarters → high-detail budgets
3–5 years → valuation, credit ratings, long-term planning
(LTP), capital structure and debt planning
A CTO's project projections should fit into this overall firm-level planning horizon — a 6-month tech proposal evaluated against a 5-year firm plan needs to explicitly show where it lands in that longer arc.
| Tool | What It Does |
|---|---|
| NPV(), IRR() | Cash flow evaluation |
| SUMPRODUCT() | Expected NPV from scenario probabilities |
| STDEV() | Volatility, for equal-probability data |
| Data → What-If Analysis → Goal Seek | Find break-even revenue or cost |
| Data Tables / Scenario Manager | Quick "what-if" tables |
You won't be asked to build valuation models from scratch — but comfort with these tools lets you sanity-check finance's output and run a quick what-if yourself before approving a tech roadmap.
┌───────────────────────────────────────────────────────────┐ │ YOU ARE NOT EXPECTED TO: │ │ ● Build full NPV/IRR/Monte Carlo models for every project │ │ ● Run daily risk analytics │ │ ● Manage FX, hedging, capital structure — that's CFO turf │ │ │ │ YOU ARE EXPECTED TO: │ │ ● Understand what NPV, IRR, sensitivity, and scenario │ │ analysis MEAN │ │ ● Understand why finance might say "this isn't financially │ │ viable" │ │ ● Read and interpret project business cases │ │ ● Challenge assumptions (revenue too optimistic? costs too │ │ low?) │ │ ● Communicate in the SAME language as the CFO and the board │ └───────────────────────────────────────────────────────────┘
The intent was never to turn a CTO into a finance manager — it was to build enough financial acumen to make better investment decisions, understand risk, and defend tech projects in financial terms.