You have reconciled the statements (lesson 6.1.1) and graded the add-backs (lesson 6.2.1). You now have your own earnings number, and it is lower than the broker's. The seller will say you are being unfair. The broker will say every buyer does this. The way to end that argument is not a better opinion. It is the seller's own paper: the contracts, invoices, customer lists and schedules that actually produce the revenue. This lesson is how to trace claimed earnings back to those documents, and why they are the most reliable check you have on the headline.
Principle
Every claimed dollar of earnings comes from somewhere: a customer, a contract, an invoice, a deposit. Trace it there, not to the deck.
The documents in a deal do not all carry the same weight. Rank them by who made them and why:
- Bank statements and tax returns. Made by third parties or filed under penalty, for reasons that have nothing to do with selling the business.
- Operating records. Contracts, invoices, customer lists, job files, payroll registers, agreement schedules. Made by the business, to run the business, before anyone planned a sale.
- Internal financial statements. Made by the owner or bookkeeper, usually for taxes and the bank.
- The marketing package. The CIM (confidential information memorandum, the broker's marketing document), the teaser, the investment deck, the adjusted EBITDA memo. Made to sell the business to you.
When two documents disagree, the one higher on the list wins. When the marketing package claims something the operating records cannot show, the claim is not supported, however confident the presentation.
There is a useful irony here. The seller's own source documents are usually more honest than the broker's summary, because they were produced for a different purpose. A footnote written by a bookkeeper for the year-end file, a pro forma built by the owner for their own planning, a contract schedule maintained to track renewals: none of these were written to win your offer. When your scrubbed number and the seller's own documents agree, the seller is arguing with their own records, not with you.
This is also not a question of whether you trust the seller. Research on acquisitions of entrepreneurial firms found that buyers' and sellers' beliefs about whether the other side was trustworthy were often wrong, in both directions (C6, 2009). Trust is not a diligence method. Documents are. And the documents that matter extend past the financial statements: a 2018 piece of practical advice to buyers was to look beyond the statements to the agreements with suppliers and landlords, and to talk to customers (JF59, 2018).
Case
Two cases, one fictional and small, one real and anonymized and larger. The method is identical.
Coastal Comfort HVAC (fictional)
The broker's memo for this fictional company presents FY2025 as "the Company's strongest year ever" and "the go-forward earnings base," and promotes "a strong relationship with a major local homebuilder that continues to refer new work" and "800-plus recurring maintenance agreements."
Now read the seller's own documents:
- The P&L footnote says FY2025 revenue includes a single commercial new-construction contract, the 41-unit Harbor Pointe townhome project, completed in the middle of the year, and that residential customers accelerated system replacements ahead of an announced equipment price increase. The broker calls FY2025 the base. The seller's bookkeeper calls it an exception.
- The job-cost report, cited in the same footnote, shows Harbor Pointe billed $930,000 and contributed about $205,000 after materials, subcontractors and field labor. Management's own estimate puts the pulled-forward replacement revenue at about $300,000. The seller's records size the spike for you.
- The balance sheet footnote says that one homebuilder, Beachline, owed about $245,000, roughly 38 percent of year-end receivables. The broker calls the relationship a strength. The seller's own balance sheet shows it is a concentration.
- Revenue by customer shows Beachline billed about $1,080,000 in FY2025, about 16 percent of revenue, against about $150,000 a year, about 3 percent, in FY2023 and FY2024. The share by revenue is under the 30 percent line of red flag RF-01. The finding is the one-project spike and the receivables.
- The tax return shows the owner took a $90,000 W-2 salary and $810,000 in distributions. The broker adds back the $90,000 and never mentions what it costs to replace someone doing the owner's job.
- In diligence, the contract request for the homebuilder's master agreement came back empty. No master agreement exists. The relationship runs on one estimator's friendship and job-by-job purchase orders.
- The maintenance agreement schedule confirmed about 810 active agreements, but 9 percent were multi-year prepayments sold at a FY2025 promotional discount. They are real, and some of their cash has already been collected by the seller.
- The bank statements tied deposits to reported revenue within 2 percent. The revenue happened.
Every adjustment that takes the broker's $1,710,000 down to roughly $1,017,000 of recurring earnings is supported by a document the seller produced, with one exception: the $225,000 market cost of a general manager, which comes from outside wage data you gather yourself. You do not need to win an argument. You need to cite page and line.
Project Marlin (real, anonymized)
Project Marlin was a cleared government-services company: roughly $25 to $30 million in revenue, about $3.5 million of adjusted EBITDA, priced at a headline multiple around 4.7 times. That is larger than most self-funded deals, but the documents and the method are the same.
Two findings came straight from comparing the marketing package to the seller's own material:
- The deck and the quality of earnings report did not agree. The investment deck's EBITDA and the QoE's EBITDA (a quality of earnings report is an accounting firm's independent test of reported profit) differed by a meaningful margin. The first question was simply: which number is the multiple priced off? If you cannot answer that in one sentence, you do not have a price yet.
- The seller's own pro forma priced in a loss the QoE ignored. Part of the revenue came from set-aside contracts, government work reserved for small or qualifying businesses, and the seller's own projection assumed that roughly half of that set-aside revenue would attrit in the first year under institutional ownership. The QoE never mentioned it. The most damaging assumption in the deal was written by the seller.
A third finding came from a source outside the seller entirely. The deck claimed a 90 percent success rate on recompetes (re-bids for expiring contracts). Public award data showed an overall win rate under 50 percent. Lesson 6.2.3 turns that check into a standard step.
Steps
1. List every earnings claim in the marketing package
Go through the CIM, teaser and adjusted EBITDA memo and write down every claim that affects earnings or their durability: the base-year figure, growth rates, recurring revenue, retention or renewal rates, customer relationships, backlog, margins. Each one becomes a row.
2. Name the source document that would prove each claim
For each row, write the operating record that would have to exist if the claim is true:
| Claim | Source document that proves it |
|---|---|
| Revenue by year | Bank deposits, tax returns, invoice register |
| Recurring revenue | Contract or agreement schedule with start, end, price and renewal history |
| Key customer relationship | Master agreement, purchase orders, two years of revenue by customer |
| Backlog | Signed contracts and remaining unbilled value, contract by contract |
| Retention or win rate | Lists of contracts up for renewal and which were kept, or public award records |
| Margin | Job-cost reports and the cost lines of the P&L |
Request the documents in one written list. What arrives late, thin or not at all is itself a finding.
3. Rebuild the claim from the source
Do not compare totals to totals. Rebuild each claim from the bottom:
- Sum the contract schedule yourself and compare it to the revenue the deck attributes to contracts.
- Take revenue by customer for two years and confirm the top customers and their share.
- Pick a sample of invoices, from large customers and small, and trace each one to a deposit in the bank statements.
- Count the agreements on the schedule and compare to the claim. Check how many are prepaid, discounted or expiring before closing.
4. Read every footnote and note
Footnotes in internal statements and tax return reconciliation notes are where a bookkeeper writes down what actually happened. Read every one, and copy any that disagree with the marketing package into your findings.
5. Find the seller's own projections
Ask for any budget, forecast or pro forma the owner has prepared for their own use, including any prepared for a lender. Compare its assumptions to the deck. Where the seller's private planning is more cautious than the public story, adopt the seller's assumption.
6. Use outside sources where they exist
Some claims can be checked without the seller: public contract award records, licensing databases, court records, customer reviews, and conversations with customers and suppliers once the seller agrees to introductions. A claim that outside data contradicts is not supported.
7. Write the reconciliation and send it
For each claim, record one of three outcomes: supported (the source document confirms it), adjusted (the source supports a smaller figure; state it), or unsupported (no source exists). Send the adjusted and unsupported items to the seller in writing, each citing the seller's own document by page and line. Price only what is supported. If a claim that the price depends on stays unsupported, you are being asked to pay for it on faith, and under a personal guarantee that is a reason to walk, not a reason to split the difference.
Deliverable
A Source-Document Tie-Out for the deal, with:
- One row per earnings claim from the marketing package: the claim, the source document requested, whether it arrived, what it shows, and the outcome (supported, adjusted or unsupported).
- The dollar effect of each adjusted or unsupported claim on your adjusted EBITDA, carried into your profit bridge from lesson 6.2.1.
- The seller's own footnotes and projections that contradict the headline, quoted with page and line.
- The written request list and the seller's responses, dated.
This tie-out is what you hand a quality of earnings provider (lesson 6.3.2) and what you bring to any price conversation. It turns "your number is too low" into "here is your document."
This lesson is education, not accounting, tax or legal advice. A licensed CPA should review the financials of any business you intend to buy.