How searchers can stop seller assumptions from quietly becoming investment facts

By Martin R. Kelwick, Author, Don’t Buy the Wrong Business

To make it easier for you to read this article, we suggest downloading it in PDF Format.

The core idea

A diligence process should not merely update the acquisition model. It should preserve the lineage of every assumption that can materially change price, earnings, cash requirements, transferability or the next commitment.

A searcher hears that the owner works about ten hours a week. The CIM says no customer represents more than 10% of revenue. The seller’s earnings bridge contains $142,000 of add-backs. The model is updated, the opportunity is discussed with investors, and the deal advances.

Weeks later, those statements may appear in an investment memo as clean inputs: low owner dependency, diversified customers, $550,000 of seller discretionary earnings. Nothing dishonest has happened. The problem is subtler: the assumptions have lost their provenance.

That is a control failure. In a small-business acquisition, the question is not only whether a number is plausible. It is whether the buyer can show where it came from, what was tested, what changed, what remains uncertain and which conclusions require professional determination.

The compression problem

An acquisition thesis is built through compression. Seller conversations become notes. Notes and financial statements become a model. The model becomes an LOI case. Diligence findings become an investment memo. Each stage necessarily reduces complexity.

Compression is useful, but it creates a predictable risk: a representation can move from something the seller said to something the deal team knows without a visible step showing why that change in status was justified.

The answer is not a longer memo. It is a better control layer behind the memo: an Assumption-to-Evidence Register.

A worked case: when 3.00x becomes 3.88x

Consider a fictional acquisition case used here only to demonstrate the method. A Texas commercial HVAC service and maintenance business is offered at $1.65 million. The seller presents $408,000 of operating earnings and $142,000 of proposed add-backs, producing $550,000 of seller SDE.

Economic item	Seller / initial case	Buyer evidence case
Asking price	$1,650,000	$1,650,000
Earnings base	$550,000 seller SDE	$425,000 working hypothesis
Price / earnings	3.00x	3.88x

Illustrative case. The $425,000 figure is a buyer-side working hypothesis, not an audit, QoE conclusion, valuation opinion or lender determination.

The price did not move. The denominator did. Relative to the buyer’s working earnings case, the acquisition now appears approximately 29% more expensive than under the seller SDE presentation.

But the disciplined response is not to announce that $425,000 is the ‘true’ earnings number. The working case itself has an evidentiary status. Some components may be supported by records; others may be estimates; still others may require an accountant, QoE provider, lender or another specialist to determine.

A strong diligence process therefore preserves two things simultaneously: the economic change and the epistemic status of the number that caused it.

Five states of knowledge

CLAIMED

A representation has been made. Record it accurately, but do not upgrade it merely because it appears in a CIM, spreadsheet or management conversation.

OBSERVED

The buyer has identified an underlying condition in source material or operations. Observation is stronger than representation, but it may not establish the economic conclusion.

SUPPORTED

Available evidence is sufficient for a defined working conclusion, with the relevant period, entity, population and definition aligned.

PROFESSIONALLY DETERMINED

A qualified professional has reached a conclusion within that professional’s scope – for example accounting treatment, legal enforceability, tax treatment or lender underwriting.

UNRESOLVED

The available evidence does not yet justify a determination. This is not failed diligence. It is a controlled state of uncertainty.

The distinction matters because spreadsheets are poor at preserving uncertainty. Once a number occupies a cell, downstream users tend to treat it as a fact. A visible UNRESOLVED can therefore be a stronger control than an unsupported estimate.

The Assumption-to-Evidence Register

For each assumption capable of changing the investment thesis, maintain a compact record with the following fields:

#	Field	Control question
1	Claim	What exactly was represented, by whom, and when?
2	Source evidence	Which records, interviews or observations bear on the claim?
3	Test	What reconciliation or procedure was actually performed?
4	Variance / contradiction	What differs from the original representation?
5	Evidence status	Claimed, observed, supported, professionally determined, or unresolved?
6	Economic consequence	Could the finding affect revenue, earnings, cash, NWC, capex, transition or price?
7	Unresolved question	What still has to be known before the next commitment?
8	Professional escalation	Who is qualified to determine what the buyer should not determine alone?
9	Thesis impact	What changed in the model, memo, negotiation or decision – and why?

The register is not a substitute for the diligence request list or the model. It is the lineage connecting them.

What this changes in practice

1. Customer concentration: preserve the contradiction, not just the corrected percentage.

In the fictional case, the initial representation is that the largest customer is below 10% of revenue. Customer-level evidence instead shows the largest customer at approximately 14.8%, with the top five at 42.0%.

The memo should not simply replace ‘<10%’ with ‘14.8%’ and move on. The contradiction itself matters. It tells the investor which representation changed, which evidence changed it, and which economic questions remain open – including retention, relationship ownership and transferability.

2. Recurring revenue: definitions can be more important than arithmetic.

A seller may describe roughly 60% of revenue as recurring. In the worked case, the evidence case separates 31% contracted maintenance, 24% repeat non-contracted revenue, and 45% project, replacement or one-off revenue.

Those categories are not economically interchangeable. Repeat customers can be valuable, but repeat behavior is not the same contractual protection as recurring revenue. The register forces the deal team to preserve the definition used rather than allowing a convenient label to survive into the investment case.

3. Owner dependency: hours are not functions.

An owner-hours estimate can create false precision. If the owner controls pricing, key customer relationships or operational knowledge, the relevant question is not simply whether the owner works ten, twenty or forty hours. It is which functions must continue after closing, who can perform them, and at what capacity and cost.

In the fictional case, exact owner hours remain unresolved while material owner functions are observed and a $105,000 replacement-capacity hypothesis is modeled. Keeping those statements separate prevents a hypothesis from masquerading as a measured fact.

4. Working capital: a point-in-time number is not a normalized requirement.

The case shows $200,000 of observed operating net working capital at a point in time. Normal NWC and peak NWC remain unresolved. That distinction is critical. A balance-sheet snapshot may be observable while seasonality, closing mechanics, collectible receivables and the appropriate peg still require further evidence and professional analysis.

A better interface with professional diligence

The framework is deliberately designed to make the searcher less likely to overreach.

A searcher should be able to identify that an add-back lacks support, that customer concentration differs from a seller representation, or that a working-capital question remains open. That does not make the searcher the QoE provider, accountant, attorney, tax adviser, lender, valuation professional or technical specialist.

The register improves the handoff. Instead of asking a professional to ‘check the deal,’ the buyer can present a traceable question: Here is the claim. Here is the source. Here is the variance we observed. Here is the potential economic consequence. Here is what remains unresolved. What is your determination within your scope?

That is a more efficient use of professional judgment – and a better defense against accidental amateur underwriting.

What investors and ICs should ask for

The investment committee does not need the entire data room reproduced in the memo. It needs confidence that material conclusions can be traced back to evidence.

• Where did this assumption originate?

• What evidence changed or supported it?

• What remains unresolved or dependent on professional determination?

• What would happen to the economics or decision if the assumption is wrong?

This also creates a useful discipline when the deal is moving quickly. The team can distinguish between unknown but immaterial, unknown and decision-relevant, and unknown that must be resolved before the next commitment.

The purpose is not to eliminate uncertainty. Private-company acquisitions will always contain uncertainty. The purpose is to stop uncertainty from disappearing simply because the deal has advanced another stage.

The decision rule

The most dangerous diligence error is not necessarily missing a document. It is losing the distinction between a representation, an observation, a supported conclusion and a professional determination.

A good deal thesis should therefore be auditable in the practical sense: another decision-maker should be able to follow the path from the original claim to the current conclusion and see every material change along the way.

A model tells you what the deal looks like if its inputs are true.
An evidence trail tells you why you are entitled to use those inputs at all.

For searchers, that means preserving the assumption-to-evidence trail as diligently as the model itself. For investors, it means asking not only, ‘What do you believe?’ but also, ‘What evidence earns that belief its place in the investment thesis?’

Author note

Martin R. Kelwick writes practical decision-making tools for entrepreneurs, business buyers and operators, with a focus on structured, evidence-based acquisition analysis. He is the author of Don’t Buy the Wrong Business: The Evidence-Based Due Diligence Workbook for Buying a Small Business.

Editorial disclosure: The Summit Commercial Services LLC transaction and all figures in the worked example are fictional and are used solely to demonstrate the methodology. The article is educational and does not constitute accounting, legal, tax, valuation, lending or investment advice.