The prior piece in this series walked through the mechanics of a leveraged buyout: how the deal gets financed, how debt gets paid down through the cash flow sweep, and how the exit produces a return on the equity invested. That model, run once, produces one number — a single MOIC, a single IRR. On its own, that number is close to meaningless, because it depends entirely on assumptions that were picked, not observed. EBITDA growth, the exit multiple, how much cash actually gets swept to debt each year — every one of those is a guess about the future, dressed up as a formula.
The fix isn’t a better guess. It’s running the same model multiple times, under different assumptions, to see how much the outcome actually depends on each one. That’s what sensitivity analysis is — and it’s the difference between a model that tells you what you want to hear and one that tells you what you need to know.
Why the Base Case Alone Is Dangerous
Every LBO model has a base case: the assumptions the buyer actually believes are most likely — steady revenue growth, stable margins, debt paydown on schedule, an exit multiple in line with entry. The base case is necessary. It’s also, on its own, the most dangerous number in the model, because a single output invites false confidence. A 27% IRR looks like a fact. It’s actually the output of five or six assumptions, any one of which could be wrong in either direction.
The discipline that fixes this is building at least two more cases alongside the base case: a downside case, where the assumptions that matter most are pushed in the unfavorable direction, and often an upside case, where they’re pushed favorably. The point of the downside case specifically is to answer one question: if the business underperforms in a plausible, not catastrophic, way — does the deal still work, or does the return (or the equity itself) disappear?
The Assumptions That Actually Move the Outcome
Not every input matters equally. Three assumptions tend to drive most of the swing in a real LBO’s return:
EBITDA growth. This compounds every year of the hold period, so a modest change in the annual growth rate produces a large difference in exit-year EBITDA by year five, which flows straight into exit enterprise value.
Exit multiple. Because the deal’s entire equity value at exit gets multiplied by this number, a change of even half a turn (say, 8.0x to 7.5x) can swing the return meaningfully — and unlike EBITDA growth, this one isn’t really within the buyer’s control at all. It’s a market assumption, not an operating one.
Leverage and cash sweep efficiency. How much debt gets used at entry, and how reliably free cash flow actually gets swept to paydown each year rather than absorbed by working capital swings or unplanned capital expenditure, determines how much debt is actually gone by exit — which is what the equity value at exit is measured against.
Other inputs matter too — interest rates on the debt, transaction fees, the hold period length — but these three account for most of the real variance in outcomes, and they’re the ones worth building explicit cases around rather than leaving fixed.
Building the Downside Case
Using the same deal from the mechanics piece — an $80M purchase (as shown in the sources and uses table there), $27M of equity invested, $55M of debt at entry — here’s the base case next to a plausible downside case.
Base case (from the prior piece):
- EBITDA grows from $10.0M to $14.0M over 5 years
- Debt paid down from $55.0M to $20.0M
- Exit multiple: 8.0x (same as entry)
- Exit Enterprise Value: $14.0M × 8.0x = $112.0M
- Exit Equity Value: $112.0M − $20.0M = $92.0M
- MOIC ≈ 3.4x, IRR ≈ 27.7%
Downside case — EBITDA growth roughly half the base case, exit multiple compressed, less cash actually swept to debt:
- EBITDA grows from $10.0M to only $12.0M over 5 years (slower growth)
- Debt only paid down to $32.0M (less free cash flow than assumed — some absorbed by working capital)
- Exit multiple: 7.0x (a full turn of compression versus entry)
Downside calculation:
Exit Enterprise Value = $12.0M × 7.0x = $84.0M
Exit Equity Value = $84.0M − $32.0M = $52.0M
MOIC = $52.0M / $27.0M ≈ 1.9x
IRR = (1.9)^(1/5) − 1 ≈ 13.7%
Reading it: the deal still returns positive money in the downside case — MOIC drops from 3.4x to 1.9x, IRR from roughly 28% to roughly 14%. That’s a large swing, but it’s not a wipeout. A downside case like this is actually a reasonably healthy sign for a deal: it means the return isn’t purely dependent on everything going right.
When the Downside Case Breaks the Deal
The version of this exercise that actually matters is when the downside case isn’t just worse — it’s bad enough that the equity holders lose money, or the debt itself is at risk of not being serviceable.
That happens through a specific mechanism worth naming directly: leverage cuts both ways, symmetrically. The same structure that turned a modest EBITDA improvement into a 3.4x equity return in the base case will turn a modest EBITDA decline into a much larger proportional loss on the equity. If EBITDA falls enough that the company can’t generate the cash flow to service interest payments — not just slower paydown, but an actual shortfall — the debt schedule from the mechanics piece stops being an assumption problem and becomes a solvency problem. At that point the equity isn’t diminished. It’s at risk of being wiped out entirely, because debt holders get paid before equity holders in every scenario, including bad ones.
This is the honest reason sensitivity analysis matters more in a leveraged deal than in an all-equity purchase of the same company. Leverage doesn’t just amplify the upside case. It amplifies the downside case by the same mechanism, and a model that only shows the base case is hiding exactly the scenario that matters most for understanding what’s actually being risked.
What a Real Sensitivity Table Looks Like
Rather than building just one downside case, the more rigorous version runs the model across a grid — commonly EBITDA growth rate on one axis, exit multiple on the other — and reports the resulting IRR at every combination.
| Exit Multiple → / EBITDA CAGR ↓ | 7.0x | 7.5x | 8.0x | 8.5x |
|---|---|---|---|---|
| 4% | 13.7% | 17.9% | 21.7% | 25.2% |
| 6% (base case growth) | 18.4% | 22.8% | 26.9% | 30.6% |
| 8% | 23.1% | 27.7% | 31.9% | 35.9% |
(Figures illustrative, built to show the shape of the table rather than sourced deal data.)
Reading a table like this tells you something a single IRR never can: how sensitive the deal actually is to the assumption the buyer has the least control over — the exit multiple — versus the one they have the most control over — operating growth. A deal that only clears an acceptable return in the top-right corner of a table like this is a deal that depends on the market being generous at exit. A deal that clears an acceptable return across most of the table is a deal that would work even if the operating story is only okay and the exit market is lukewarm. That’s the difference between a resilient deal and a deal that’s really a bet on multiple expansion wearing an operating thesis as a disguise.
Explain It Like I’m Four
Imagine you’re planning a lemonade stand for the summer, and you’re trying to figure out if it’s worth doing.
You could just guess: “I’ll probably sell 100 cups a day, and it’ll probably be sunny most days.” Based on that one guess, you decide it’ll definitely be worth it. But that’s just one guess dressed up to look like a plan.
A smarter way is to ask three different questions instead of one. What if it’s a great summer — lots of sunny days, lots of customers? What if it’s an okay summer — some sun, some rain, an average number of customers? And, the important one: what if it’s a bad summer — more rain than usual, fewer people walking by?
If the lemonade stand still makes you some money even in the “bad summer” version, that’s a real plan — it works even when things don’t go your way. But if the bad-summer version means you actually lose money on cups and lemons you already bought, that’s important to know before summer starts, not after.
There’s one more thing that makes this extra important for a lemonade stand you built mostly with borrowed money instead of your own. If you borrowed money from a friend to build the stand, you have to pay your friend back first, no matter how the summer goes — good or bad. So a bad summer doesn’t just mean less profit for you. It can mean there’s nothing left over for you at all, even though your friend still gets paid.
One sentence version: don’t just plan for the summer you’re hoping for — plan for the summer that might actually happen, especially if some of what you spent was borrowed and has to be paid back no matter what.
What’s Next
The remaining piece in this thread is where the leverage itself creates risk that has nothing to do with any single assumption being wrong — the mechanical way debt amplifies both outcomes, and how that connects to the power law and information-edge thinking running through the rest of this site.
Where in your own numbers have you only ever modeled the version where things go right?
About Me
I'm Michael Philippou, co-founder of Big Love, a plant-based ice cream business in Santa Monica that I've been running with my wife Victoria for over ten years. Before ice cream, I was a lawyer. I write about the real, unpolished lessons of running a small business — no gurus, no hype, just what's actually worked and what hasn't. Watch more on YouTube (Real Business Real Lessons) or follow along on LinkedIn.
Want to know how your business is actually doing? Start with the free Business Health Check — a couple minutes of questions and your email gets you a real score. Or dig into the other four: Valuation Estimator, Cash Runway, Customer Concentration Risk Checker, and Break-Even Calculator.