Sensitivity analysis example: stress-testing an $8.4 billion take-private

The LBO model of Clearwater Analytics’ take-private returns 21.1% on an outside-in case at the $24.55 offer. This page stress-tests that number on the same model: first which inputs move it and by how much, then how the four that matter trade off against each other in two-variable grids.

The deal turns on growth. The IRR stays above 20% only while revenue growth averages at least 11.5% a year from 2027 to 2030; the outside-in case assumes 12.1%, a cushion of 0.6 points a year.

Which inputs move the return

Revenue growth moves the IRR most: 2 points a year either way spans 17.5% to 24.6%. The exit multiple and the margin follow; SOFR, merger costs, capex, working capital, minimum cash and the PIK election each move it 0.6 points or less.

Each bar runs one input from the low to the high end of its range with every other input at base, and reads the FY2030 IRR at each end. The bars are sorted by the width of that span, so the chart reads as a ranking: the top three drivers decide the deal, entry price matters at the margin, and the bottom seven each move it less than a point across their full range.

The ranking is only as good as the ranges, so each comes from Clearwater’s own record rather than a uniform ±10%. A flat ±10% would flex SOFR by 35 basis points and the exit multiple by 2.5 turns, which says more about the starting values than about the risk. The basis column gives the reason for each range.

Why growth leads. Two points of growth a year compound over the four forecast years to about 7% more revenue in 2030, and the exit is priced on next-year EBITDA, so the gap carries into 2031 as well. Higher growth also produces more free cash flow along the way, which pays down debt. All three effects land on the equity at exit.

Why the exit multiple and the margin follow. Two turns on a 25.0x exit move enterprise value by about 8%. Net debt at exit is about a sixth of enterprise value, so the equity moves by a little more than that; across the 4-turn range the IRR spans 5 points. Two points of margin on a 38.8% base change 2030 EBITDA by about 5%, a smaller lever than two points of growth a year.

Why entry price matters less than it looks. A dollar a share is about 4% of the price, but the debt is sized on EBITDA, not on price, so every extra dollar is funded with equity. That raises the equity check by about 5% and moves the IRR by about 3 points across the $2 range.

Why the rest barely matter. Stock-based compensation is the largest of the small items: the model treats all $128 million as a cash cost, and if half moved into a sponsor equity pool the IRR would be 21.8%. Merger costs are paid once at close. SOFR, capex, working capital and minimum cash change the cash available to repay debt by tens of millions a year against an equity value in the billions. The PIK election only defers part of the margin until June 2028, and the deferred interest is repaid at exit.

Entry price against exit multiple

At the $24.55 offer, a 20% IRR needs the exit multiple to hold near the 25.0x paid: 23.0x gives 18.5% and 27.0x gives 23.5%. At $22.00 the deal would clear 20% even with a 2-turn de-rating, at 22.4%.

This is the grid an investment committee reads first, because it sets the two numbers the buyer controls or bets on: the price paid and the price received. Read across a row to see what overpaying costs at a given exit; read down a column to see how much multiple expansion a given price needs. Each $1.25 a share moves the IRR by about 1.6 to 2.0 points, and each turn of exit multiple by about 1.2 to 1.3 points. The green cells run diagonally: a higher price is affordable only with a higher exit. At $27.00 it takes a 27.0x exit, two turns of expansion, to clear 20%, at 20.2%.

Exit multiple against revenue growth

Two points less growth a year takes the IRR to 17.5% at a 25.0x exit; recovering 20% then needs an exit above 27x. Two points more growth clears 20% even at a 23.0x exit, at 22.0%.

This grid pairs the two largest drivers. Growth columns show the average for 2027 to 2030, with each year shifted by the same points. Growth and the exit multiple are linked in practice, since buyers pay more for faster-growing software, so the top-left and bottom-right corners are the likelier pairings. A business growing 8% a year is unlikely to keep a 25.0x multiple: at 25.0x it returns 13.8%, and at 21.0x 8.4%. Each two points of average growth is worth about 3.5 points of IRR at any exit multiple, against about 2.5 points for each two turns of multiple.

Exit multiple against EBITDA margin

Margin is a weaker lever than growth: two points below plan returns 19.2% at a 25.0x exit, and a 27.0x exit takes it back above 20%, at 21.6%. At a 23.0x exit, it takes two points of extra margin to clear 20%, at 20.3%.

The margin columns show the 2030 adjusted EBITDA margin, with every forecast year shifted by the same points. The case reaches 38.8% in 2030, which assumes Clearwater keeps expanding margin at close to its 2021 to 2025 pace. Each two points of margin is worth about 1.7 to 2.0 points of IRR, against about 2.5 points for each two turns of exit multiple, so the multiple decides the return more than cost control does.

Revenue growth against EBITDA margin

Missing on both costs more than either alone: two points less growth and two points less margin take the IRR to 15.6%. Two points of extra margin does not offset two points of lost growth, at 19.3%; it takes four, for just over 21%.

This grid holds the exit at 25.0x and isolates the operating case. Read along the diagonal from top-left to bottom-right to see a uniform miss or beat on both; the corners span 9.8% to 31.6%. Growth is the steeper axis: moving one column changes the IRR by about 1.9 points, and moving one row by about 3.5. A plan that trades growth for margin, for example by cutting sales spend, has to find about two points of margin for each point of growth it gives up to hold the return.

What to watch. Organic growth each quarter against the 11.5% a year the case needs; the multiple buyers pay for vertical software when the hold ends; and the margin from the Enfusion and Beacon integration. The financing terms, SOFR and the PIK election included, barely move the return; the debt schedule covers what they do to leverage and coverage.

Methodology. Every figure is from a copy of Compound’s LBO model with a Sensitivities tab added; the base ties to the LBO model at 21.1% and 2.37x, and the outside-in case is described on the LBO page. Each grid is an Excel data table on the model, so it recalculates when an assumption changes. Excel reads a data table’s row and column input cells only from the table’s own sheet, so each driver has an input cell beside its grid that the model references, and calculation is set to automatic with data tables included. The driver ranking is one more data table that runs each driver at its low and high end. The model was built in Compound; it rebuilds work a deal team could have done and is not the buyers’ model.

FAQ