What an LBO Model Is Actually For
It is worth stating plainly: an LBO model is not a forecasting exercise. Its job is to answer one question with precision — at this price, with this capital structure, under this set of operating assumptions, what return does the equity earn, and how much does that return degrade as reality deviates from plan?
Everything in a well-built LBO model serves that question. The three-statement projections exist to drive free cash flow, which drives debt paydown, which drives the equity check at exit. The covenant schedule exists to identify the point at which the deal breaks before it ever gets the chance to return capital. The sensitivity tables exist to show the IC the entire distribution of outcomes, not just the headline case the deal team is excited about.
A model that nails the base case mechanics but cannot answer "what if exit multiple compresses 1.5 turns and EBITDA growth is half of plan" has not actually done its job — it has produced a number, not an analysis.
The Core Mechanics, Briefly
This is not a primer on LBO mechanics — if you need that, there are hundreds of resources. The short version, for context on everything that follows:
- Sources and uses sets the entry capital structure: purchase price, debt tranches (term loan, revolver, possibly mezzanine or unitranche), and the sponsor equity check that fills the gap.
- The operating model projects revenue, margin, and free cash flow forward, typically 5–7 years, feeding a debt schedule that sweeps excess cash to pay down principal.
- The exit applies an assumed multiple to exit-year EBITDA, nets out remaining debt, and the resulting equity value, divided by the entry equity check, gives MOIC and IRR.
The mechanics are well understood across the industry. What separates a mediocre LBO model from an excellent one is almost never the formula logic — it is the scenario and sensitivity architecture wrapped around that logic.
Building the Scenario Stack
A serious LBO model carries, at minimum, four named scenarios. Each must run through the identical model structure — the only difference between them should be the assumption inputs, never the calculation logic.
Management case
The projections provided by the target company's management team, typically the most optimistic of the set. Sponsors model this separately from their own base case specifically so the IC can see the gap between what management is projecting and what the deal team actually underwrites to — a gap the IC will always ask about.
Sponsor base case
The deal team's own underwriting case, usually a haircut on management's revenue growth and margin expansion assumptions. This is the case the investment thesis is actually built on, and the one the IC will scrutinize most closely for realism.
Downside case
A genuinely conservative case — not a token haircut, but a scenario that reflects what happens if the macro environment deteriorates, a key customer is lost, or margin expansion simply doesn't materialize. The discipline here is resisting the urge to make the downside case mild enough that it still shows an acceptable return. A downside case that always clears the fund's hurdle rate is not doing its job.
Covenant stress case
Distinct from the downside case in purpose: this scenario exists specifically to find the point at which the deal breaches its credit agreement covenants — typically a leverage ratio (Net Debt / EBITDA) or a fixed charge coverage ratio. It answers a different question than the downside case: not "does the equity still return capital," but "does the company stay solvent and in compliance with its lenders."
Why these must share one model: If the downside case is built in a separate file from the base case, any later change to the base case — a revised purchase price, an updated debt quote, a corrected formula — will not propagate to the downside case unless someone remembers to copy it over by hand. By the time the IC memo is finalized, it is common for the "downside case" to actually reflect a stale version of the base case's model logic. This is the single most common integrity failure in LBO models that have been through multiple revision rounds.
The Sensitivity Tables Every LBO Model Needs
Beyond the four named scenarios, the IC will expect to see the full sensitivity space — not just discrete cases, but a continuous view of how returns move as key assumptions vary.
Entry multiple × exit multiple
The single most important table in any LBO model. Entry multiple on one axis, exit multiple on the other, IRR or MOIC in the grid. This table answers the question every IC member is implicitly asking: "What if we pay too much, or sell into a worse market than today's?" A common convention is to show entry multiples from 0.5x below to 0.5x above the bid price, and exit multiples spanning a full turn below the entry multiple (no multiple expansion) up to a turn above it.
Leverage × exit multiple
Shows how returns change with more or less debt at entry. Higher leverage amplifies equity returns in the base case but also amplifies downside risk and tightens covenant headroom — this table makes that tradeoff explicit for the IC and the financing committee.
Revenue growth × margin expansion
The operational sensitivity. This table is less about the deal structure and more about execution risk: how much of the return depends on the company actually growing and expanding margins as projected, versus how much return exists even with flat operating performance (i.e., from deleveraging and multiple arbitrage alone).
Hold period × exit multiple
Tests whether the deal still clears the fund's return hurdles if the exit is delayed — a real risk in periods of soft M&A or IPO markets. A deal that only works on a 4-year hold and falls apart at 6 years carries meaningfully different risk than one that holds up across the range.
Covenant Stress Testing: The LBO-Specific Discipline
Covenant analysis is the part of LBO modeling that gets the least attention in generic financial modeling content, because it is specific to leveraged structures. It deserves its own section because getting it wrong has real consequences — a deal that looks attractive on a returns basis can still be uninvestable if it breaches covenants under plausible stress.
What to model
The credit agreement will specify the actual covenants — commonly a maximum total leverage ratio (Net Debt / EBITDA) and sometimes a minimum fixed charge coverage ratio (EBITDA – Capex / Debt Service). Build both as a running calculation throughout the projection period, not just at exit.
Run it under the downside case, not the base case
A common error is checking covenant compliance only under the base case, where it almost always passes trivially. The real question is whether the covenant holds under the downside case — and at what point in the downside trajectory a breach would occur. If the covenant breaches in year 2 of a 5-year downside case, that is materially different information than a breach in year 5, because it changes how much operating runway the company has before lenders can call a default.
Build an EBITDA cushion sensitivity
A useful exhibit for the IC: at each point in the projection, how far can EBITDA fall (holding debt and covenants fixed) before a breach occurs? This "covenant cushion" percentage is one of the most informative single numbers in a leveraged deal — it tells the IC, in plain terms, how much operating underperformance the capital structure can absorb.
Common mistake: Modeling covenant compliance as a single point-in-time check rather than a running calculation across the full hold period. Leverage ratios typically step down on a schedule in credit agreements (e.g., 6.0x in year 1, stepping to 5.0x by year 3) — a model that doesn't track the covenant schedule against the actual projected ratio at every period will miss breaches that the static check doesn't catch.
What Separates a Good LBO Model From a Great One
Every output is traceable to a specific input
If the IC asks "what's driving the 3-point IRR gap between the base case and the downside case," the answer should be immediate and precise: "1.8 points from the lower exit multiple, 0.9 points from slower deleveraging due to lower EBITDA, 0.3 points from the higher implied entry leverage on lower EBITDA at close." A model where that breakdown requires twenty minutes of cell-tracing is not presentation-ready, regardless of how sophisticated its mechanics are.
Scenarios are structurally guaranteed to share the same logic
As covered above — if scenarios are separate files or separate sheets that were copied at some point in the past, they will drift. A model where the base case and downside case are guaranteed by construction to share the same revenue build, the same debt schedule logic, and the same exit mechanics is the only way to ensure a comparison between them is actually meaningful.
The presentation layer is built, not improvised the night before the IC
The best modelers build the output exhibits — the sensitivity tables, the covenant cushion chart, the scenario comparison bridge — as part of the model itself, refreshing automatically when inputs change. The alternative, common at firms without strong modeling discipline, is rebuilding these exhibits manually in PowerPoint the night before the IC meeting, with numbers that can silently fall out of sync with the underlying model.
The model survives diligence updates without rework
Diligence will produce updated information — a corrected working capital assumption, a revised management forecast, a different quote on the debt financing. A well-architected model absorbs these updates by changing inputs in one place and watching every scenario and sensitivity table update automatically. A poorly architected one requires the analyst to manually propagate the change across four scenario tabs and a dozen sensitivity tables, with high risk of missing one.
How Top Firms Approach This
Practices vary, but the discipline at the top of the industry is consistent on the core principle: scenarios should never be separate files. Firms like KKR, Apollo, Blackstone, and Bain Capital — and the investment banks that support their deals, including Goldman Sachs' and Morgan Stanley's leveraged finance and sponsor coverage groups — build internal modeling standards specifically to prevent the scenario drift problem described above.
The common solution is a structured scenario-selector architecture: a single input block where each column represents a named scenario, and a master switch feeds the active column's values into the calculation engine. This is a meaningful improvement over file-copy scenario management, but it has its own ceiling — every new assumption requires populating a value in every scenario column, and the architecture becomes unwieldy past 5–6 scenarios or when sensitivities need to be layered within scenarios.
The newer approach, increasingly used by analysts who have grown frustrated with both file-copy chaos and scenario-selector rigidity, is true branching: each scenario is defined only by what it overrides relative to the base case, inheriting everything else automatically. A covenant stress case becomes "downside case, but with leverage 0.5x higher and no revolver headroom" — three explicit overrides, not a fully rebuilt model.
Common LBO Modeling Mistakes
| Mistake | Why it matters |
|---|---|
| Circular reference instability | Interest expense depends on the debt balance, which depends on the cash sweep, which depends on free cash flow after interest — a genuine circularity. Without proper iterative calculation settings and a circuit breaker, the model can produce unstable or incorrect results under certain scenarios without any visible error. |
| Hardcoded exit multiple equal to entry multiple | A model that assumes no multiple change at exit is not stress-testing anything — it is hiding the single most volatile assumption in the entire analysis behind an unstated constant. |
| Downside case built as a token haircut | If every named scenario still clears the fund's hurdle rate, the scenario set is not providing real information to the IC — it is theater. |
| Covenant check only at one point in time | Misses interim breaches that occur before the final exit year, which is exactly when a deal is most likely to actually get into covenant trouble. |
| Management fees and carry left out of the equity waterfall | Gross IRR looks materially different from net IRR once fund economics are applied — an LP-facing exhibit that only shows gross returns is incomplete. |
| Scenario tabs that have silently diverged | The most common integrity failure in models that have been through multiple revision cycles — see the callout above. |
The Exhibit Set a Strong IC Memo Needs
Beyond the model itself, the output exhibits that should accompany a well-built LBO model:
- Returns summary table across all named scenarios — MOIC and IRR for management case, base case, downside, and stress case side by side.
- Entry/exit multiple sensitivity grid — the standard 2D table described above.
- Sources and uses — capital structure at entry, clean and simple.
- Leverage and coverage ratio chart — running through the hold period under the base case and downside case, with covenant thresholds overlaid so breaches are visually obvious.
- Value creation bridge — decomposing the equity return into EBITDA growth, multiple expansion/contraction, and deleveraging, so the IC can see exactly what the return depends on.
- Downside-to-base bridge — the specific assumption deltas that separate the downside case from the base case, quantified individually.
Every one of these exhibits should regenerate automatically from the model when inputs change. If any of them require manual updates in a separate document, they will eventually fall out of sync with the model that drives the actual investment decision.
TreBranch is Git for spreadsheets
Build your base case once. Branch the management case, downside, and covenant stress case off it — each inherits everything and overrides only what changes. Right-click any input to run an entry/exit or leverage sensitivity sweep in seconds. DiffPanel shows the IC exactly what's driving the gap between any two scenarios, automatically. No more scenario tabs quietly drifting apart before the memo is due.
Try Free — Microsoft Store →7-day free trial. Buy once, own it forever. 100% offline — your models never leave your computer.
Frequently Asked Questions
What is a good IRR for an LBO?
Most PE funds underwrite to a gross IRR target in the 20–25% range at the deal level, reflecting the net returns LPs expect after fees and carry (typically targeting net IRR in the mid-to-high teens). The required IRR varies by fund strategy, deal size, and risk profile — growth equity and venture-adjacent buyout strategies often underwrite to higher targets given the additional execution risk, while infrastructure-adjacent or lower-leverage deals may underwrite to somewhat lower returns given the reduced risk.
How many scenarios should an LBO model include?
Four named scenarios is the practical standard for an IC-ready model: management case, sponsor base case, downside case, and a covenant stress case. Beyond named scenarios, the model should support continuous sensitivity sweeps (entry/exit multiple, leverage, revenue growth) that go beyond what a handful of discrete scenarios can show.
What causes circular references in an LBO model, and how do you handle them?
The circularity arises because interest expense depends on the average debt balance during the period, which depends on the cash sweep amount, which depends on free cash flow after interest expense — a genuine loop. Excel handles this via iterative calculation (File → Options → Formulas → Enable iterative calculation), but a more robust practice is to add a circularity "switch" — a cell that can force interest expense to zero, allowing you to break the loop intentionally to debug the model or isolate calculation errors.
What's the difference between a covenant stress case and a downside case?
A downside case answers "does the equity still earn an acceptable return if the business underperforms?" A covenant stress case answers a narrower but more urgent question: "does the company stay in compliance with its credit agreement, or does it default?" These can use overlapping assumptions, but the covenant stress case specifically needs to track the leverage and coverage ratios period by period against the actual covenant schedule in the credit agreement — a downside case focused only on equity returns can miss an interim covenant breach entirely.
How do you build a value creation bridge in an LBO model?
The standard decomposition splits the total equity value increase into three components: EBITDA growth (change in EBITDA at constant multiple), multiple change (change in multiple at constant exit-year EBITDA), and deleveraging (debt paydown over the hold period). Some practitioners add a fourth bucket for cash generation beyond debt paydown. The bridge is typically built as a waterfall chart starting at entry equity value and ending at exit equity value, with each component shown as a labeled step.
Should management fees and carried interest be included in LBO model returns?
Deal-level returns (gross IRR/MOIC) are calculated before fund economics and represent the return on the deal itself. Fund-level or LP-facing returns (net IRR/MOIC) subtract management fees (typically 1.5–2% annually) and carried interest (typically 20% of profits above a preferred return hurdle, often 8%). An IC memo evaluating a specific deal usually focuses on gross returns, since fund economics are constant across deals, but any external-facing communication to LPs should clearly distinguish gross from net.