From Summit Corp Dev — Acquisition Analyst
Method for a first-pass quality-of-earnings review of a lower-middle-market service-business acquisition target — building the adjusted-EBITDA bridge, disciplining add-backs, normalizing owner compensation, scrutinizing revenue quality, pegging normalized working capital, and handling cash-basis financials. Use when reviewing a target's historical financials to establish a defensible normalized earnings base.
How this skill is triggered — by the user, by Claude, or both
Slash command
/summit-corp-dev:qoe-normalizationThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
This skill governs the first-pass quality-of-earnings review of a target's historical financials. Its job is to establish a defensible **normalized adjusted EBITDA** — the earnings base the valuation multiple applies to — and to surface what is wrong, unverified, or unsustainable in the reported numbers.
This skill governs the first-pass quality-of-earnings review of a target's historical financials. Its job is to establish a defensible normalized adjusted EBITDA — the earnings base the valuation multiple applies to — and to surface what is wrong, unverified, or unsustainable in the reported numbers.
In a real transaction, a confirmatory quality-of-earnings audit is performed by a specialist accounting firm with access to the general ledger, bank statements, and management. This skill does not replace that. It is a first-pass internal review from the historical statements and owner-provided facts, sized to inform a screening decision and to scope what confirmatory diligence must test. Say so in the output. Reported earnings are owner-provided and unaudited unless stated otherwise — label them that way.
Target financials in this segment commonly arrive as PDFs, scanned tax returns, QuickBooks exports, or photographed pages. Fable 5's document extraction is strong — ingest these files directly and tie every figure to its page/source — but treat extraction as not the same as verification, and keep the owner-supplied/unaudited label. Tie every figure in the bridge to a specific source (file name, page, line), preserve the owner-provided / unaudited label through every downstream number, and surface input problems rather than papering over them: cash-basis books, missing months, unsegregated COVID/ESSER relief income, commingled owner expenses, route-level revenue that does not roll up cleanly to the P&L. If a figure cannot be tied to a source, mark it Unverified and ask, rather than guess.
The valuation multiple applies to normalized adjusted EBITDA, not to reported operating income or reported net income. Build the bridge explicitly, for every historical year, so the trend is visible and the most representative run-rate year can be identified.
The bridge runs:
Reported operating income
+ Depreciation & amortization (non-cash)
+ Interest (a financing item; EBITDA is pre-financing)
+/- Owner / officer compensation (normalize to market-replacement cost)
- One-time & non-operating items (remove)
+/- Related-party adjustments (normalize to market)
= Normalized adjusted EBITDA
Every line is a labeled adjustment with a rationale. Do not collapse adjustments into a single normalized number — the bridge has to be legible to a reviewer who will challenge each line.
This is usually the largest adjustment and the one most exposed to judgment. The reported owner compensation is rarely the right number, and it often swings year to year.
Do not simply delete owner compensation. Normalize it:
Build an add-back register and label every item. An add-back is defensible only if it is both non-recurring and verifiable. Everything else is scrutinize.
Be conservative. An inflated add-back inflates EBITDA, inflates the implied purchase price at any multiple, and is the first thing a confirmatory QoE will cut. Erring toward "scrutinize" protects the analysis.
Normalized EBITDA is only as good as the revenue under it. In contracted student transportation the target effectively is its contract book, so revenue-quality work is contract-book work. Assess:
student-transportation-market skill.Conflicting until segregated.Review the accounts-receivable (district A/R, often slow-pay), parts/fuel inventory, and accounts-payable trend across the historical years and propose a normalized working-capital peg — the level of working capital the business needs to operate. This matters to the deal: the buyer funds a normalized level of working capital, and a peg set wrong moves real money at close. Note any unusual movement (a receivables build tied to a slow-paying district, a parts-inventory swing tied to a fleet refresh) and whether it is operational or a timing artifact.
Examine the gross-margin trend and scrutinize the operating expense lines, with attention to the lines that swing year to year — owner compensation, driver wages and benefits, fuel, maintenance and parts, insurance, depreciation, professional fees, yard/terminal rent. A volatile expense line is either a normalization candidate or a sign of inconsistent bookkeeping; either way it needs an explanation. Pay particular attention to driver wages vs. contract escalators (the central operating squeeze in the sector) and insurance (a primary day-one synergy lever for a strategic acquirer, but in the standalone numbers it is what the seller actually paid).
School-bus operators are maximally asset-heavy: buses run roughly $110-130K diesel and ~$300-400K electric, with multi-year useful lives, and the fleet must be continuously refreshed. EBITDA alone overstates cash earnings — EBITDA minus maintenance capex is the truer cash figure. Report both, disclose the maintenance-capex assumption explicitly, and flag it as a diligence item. (Comparable signal: NEXS sold at roughly 5x EV/EBITDA but ~50x EV/EBIT — the gap is depreciation, and depreciation in this sector is real, ongoing fleet replacement, not an accounting artifact.)
If the financials are cash-basis, or the company does not perform a hard monthly close, flag it prominently as a limitation. Do not reconstruct the statements on an accrual basis. State the specific distortions to expect:
Carry this caveat into every figure that depends on the financials, and recommend an accrual-basis review as a confirmatory-diligence item. The pipeline names this risk; it does not fix it.
Surface, prominently, anything in this category — these reprice or stop a deal: district concentration (especially top-1); a declining revenue or margin trend without a clean explanation; net routes lost at recent re-bid without a strategic reason; safety/claims tail — accident frequency, at-fault trends, deduction or penalty history with districts, FMCSA rating trajectory (a final FMCSA "Unsatisfactory" / Out-of-Service Order is contract-existential); add-backs that do not hold up to scrutiny; owner dependence so deep the district relationships may not transition; related-party arrangements (owner-leased yard, owner-affiliated fuel or parts supplier) that are not at market; commingled or unsegregated relief income (ESSER, PPP, FEMA); cash-basis accounting without a hard close; and any inconsistency between the income statement, the balance sheet, and the owner's narrative.
Reporting public adverse facts (accidents, litigation, deduction history, FMCSA ratings) is legitimate corp-dev diligence — record them crisply and factually, keep caveats concise, and do not soften or omit them.
A QoE Findings Memo: earnings-quality summary with the headline normalized adjusted EBITDA and the year it rests on; the adjusted-EBITDA bridge by year; the add-back register with each item marked defensible or scrutinize; the owner-compensation normalization with components shown; the revenue-quality assessment (contract book, district concentration, bid-season net wins/losses, relief-income segregation); the normalized working-capital peg; EBITDA minus maintenance capex alongside headline EBITDA; the accounting-quality limitations with the cash-basis flag prominent if applicable; and the red-flag and diligence list. Lead with the finding; this memo is the earnings basis for the valuation and the IC model.
A step is done only when concrete artifacts exist — the bridge file is written, every figure is tied to a labeled source, the add-back register is populated, the red-flag list names specific items — not when the analysis is asserted to have been performed. Self-assertion is not completion.
A worked example (an illustrative / hypothetical mid-Atlantic student-transportation operator) is in examples.md in this skill's directory — load it if you want a concrete illustration of the bridge and add-back discipline. Treat the numbers as illustrative, not as a real Summit deal.
npx claudepluginhub brisket1994/summit-corp-dev --plugin summit-corp-devProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.