A project produces a 24% IRR. The board approves. The return exceeds the hurdle rate. The investment proceeds.
But the 24% is an output. It tells the board nothing about what produces it. And the drivers of the 24% determine whether the return is robust (built on multiple, diversified value drivers) or fragile (dependent on a single assumption that may not hold).
The IRR Tree is a decomposition methodology that breaks the aggregate return into its component drivers, identifies which variables contribute the most to the return and reveals the assumptions on which the investment’s economic viability depends.
The Decomposition
IRR = f(Revenue Growth, Margin, Cash Conversion, Exit Value)
Each component is further decomposed:
Revenue Growth = Volume Growth x Price Growth
Margin = Revenue x (1 minus Cost Ratio), where cost ratio decomposes into material cost, labour cost, overhead and SG&A
Cash Conversion = EBITDA minus Tax minus Capex minus Working Capital Change
Exit Value = Exit-Year EBITDA x Exit Multiple
The IRR Tree traces the return from the aggregate percentage down to the individual assumptions. The tree reveals which assumptions contribute the most to the return.
What the Tree Shows
Example decomposition
A PE fund evaluates a target at Rs 200 crore entry valuation with a projected 24% IRR over five years.
The tree shows:
40% of the IRR comes from exit multiple expansion (entry at 8x EBITDA, exit at 11x). 25% comes from margin expansion (from 14% to 19%). 20% comes from working-capital release (days reducing from 90 to 65). 15% comes from revenue growth (12% CAGR).
The board now knows: this is not a revenue-growth investment. It is a multiple-expansion and margin-improvement investment. The revenue growth, which the management presentation emphasises, contributes only 15% of the return. The exit multiple, which the management presentation barely mentions, contributes 40%.
The investment’s viability depends primarily on whether the exit market will pay 11x in five years. If the exit multiple stays at 8x (no expansion), the IRR drops from 24% to 14%. If margins do not expand (staying at 14%), the IRR drops to 17%. If both assumptions fail, the IRR drops to 8%, below most hurdle rates.
The tree converts a single comfortable number (24%) into a set of testable assumptions, each with its own probability and sensitivity.
The AI Capability
AI transforms the IRR Tree from a manual sensitivity exercise into a continuous analytical system.
Driver identification: AI can automatically identify which variables contribute the most to the IRR by running thousands of simulations with different input combinations and measuring each variable’s contribution to the output variation.
Interaction effects: Revenue growth and margin expansion may interact (faster growth may compress margins due to scaling costs). AI can model these interactions, which static sensitivity analysis (changing one variable at a time) misses.
Assumption testing: AI can compare the investment’s assumptions to historical performance of similar businesses, industry benchmarks and macroeconomic data. An exit-multiple assumption of 11x can be tested against the historical distribution of exit multiples in the same sector: if the median is 8x and 11x represents the 85th percentile, the assumption is aggressive.
Scenario generation: Instead of three scenarios (base, best, worst), AI can generate probability-weighted distributions for each variable and simulate thousands of combined scenarios, producing a probability distribution for the IRR rather than a point estimate.
Real-time recalibration: Post-investment, AI can continuously recalculate the IRR Tree as actual performance data replaces assumptions. The tree shows, in real time, whether the original return thesis is on track or whether the dominant driver (exit multiple, margin, growth) is diverging from plan.
In Northrop Management Private Limited’s capital allocation and transaction advisory, the IRR Tree is applied to every significant investment decision and every acquisition valuation. The tree answers the question that the aggregate IRR cannot: what are we actually betting on?
Ashish Chaudhary, frames the diagnostic power directly: “A 24% IRR is a conclusion. The IRR Tree is the reasoning. The board that sees only the conclusion approves a number. The board that sees the tree approves a thesis: this investment generates its return primarily through exit-multiple expansion and margin improvement, with revenue growth contributing only 15%. That thesis can be tested, monitored and challenged. The number cannot.”
Questions for the Boardroom
- For this investment, what percentage of the projected IRR comes from each driver (revenue growth, margin expansion, working-capital improvement, exit-multiple expansion)?
- Which single assumption contributes the most to the return, and what happens to the IRR if that assumption does not hold?
- Is the exit-multiple assumption consistent with the historical distribution of exit multiples in this sector, or does it require above-median market conditions?
- Have we modelled the interaction effects between drivers (e.g., does faster growth compress margins), or have we treated each variable independently?
- Post-investment, are we tracking actual performance against the IRR Tree’s driver assumptions, so we know which elements of the thesis are materialising and which are not?
Closing Implication
The IRR Tree is the diagnostic that converts a single return number into a map of value drivers, assumptions and sensitivities. The board that sees the tree knows what the investment depends on. The board that sees only the IRR knows what the investment promises. The difference, in governance terms, is the difference between understanding a decision and approving a calculation. # Series III: AI x Management Intelligence