Model debrief: your model shows about 31% IRR on Rs 300 crore of revenue, margin rising from 30% to 40%, entry at 20x EBITDA with 6x debt and exit at 18x in year 5. Which three assumptions drive the IRR, and what is it if the margin stays at 30%?

Margin first, exit multiple second, growth third; with a flat 30% margin the IRR drops from about 31% to about 23%. Moving margin between 30% and 45% swings IRR from 23% to 35%; the exit multiple between 14x and 22x swings it from 25% to 37%; growth between 10% and 26% swings it from 21% to 41%. Ten points of margin on Rs 686 crore of year 5 revenue is Rs 69 crore of EBITDA at 18x, and that is most of the return.

Move each one alone and watch the IRR, the way you would test which tap is leaking by closing them one at a time. Hold two of the three at base, swing the third across a plausible range, and the width of each swing is that assumption's weight. The base case has EBITDA rising from Rs 90 crore to about Rs 275 crore, exit value of about Rs 4,942 crore and debt down to about Rs 17 crore, so Rs 1,260 crore of equity becomes about Rs 4,924 crore, 3.91x. The debrief is not asking you to recite the model; it is asking whether you know which cell you would check first if the deal went wrong.

Year 5 revenue is about Rs 686 crore either way. At 40% that is Rs 275 crore of EBITDA; at 30% it is Rs 206 crore, Rs 69 crore less, which at 18x is Rs 1,235 crore of exit value gone; the IRR falls to about 23%. Less cash is generated along the way too, so debt at exit is about Rs 153 crore instead of Rs 17 crore. The return does not vanish, because 18% growth at 18x still works, but it goes from a top-quartile deal to an ordinary one on a single assumption, and that is the sentence the interviewer is waiting for.

That the margin plan is the deal. Ten points of margin on a software company means cutting sales and marketing as a share of revenue, raising prices, or pushing gross margin up by moving customers to cloud; each is a different plan with different evidence, and the memo should say which one and show the cost lines that move. The exit multiple is second: a 18x exit on a business that has just finished expanding margin assumes a buyer who still sees growth, so the model should show 14x as a downside, not a stress. Growth is third, which is unusual for software and is a direct result of entering at 20x, where the price already pays for the growth. Say the limit: the tornado moves one input at a time, and in a real downturn margin, multiple and growth fall together.