Operating Data Versus the Management Presentation

A management presentation is the best account of the business the people selling it can assemble. That is the document doing its job, not a criticism of it. Confirmatory diligence exists because the buyer has a different job: to find out whether the systems that ran the business recorded what the slides describe, and to do it before signing, while the answer can still change the deal.

The Presentation Is Built to Sell, Not to Prove

Every figure on a slide arrives already framed. A definition was chosen, a period was chosen, a population was chosen, and none of those choices is visible until you ask. Nothing improper about that. It means the deck is an argument, and an argument is not a record.

Auditing practice is blunt about where that leaves a buyer. Asking company personnel and accepting the answer does not amount to sufficient evidence on its own. Inquiry opens a line of work. It does not close one. A deal team that treats a management statement as a finding has skipped the step that separates the two. Observed behavior is the check on stated behavior: what customers and teams actually did, rather than what they said they did.

Why the Agreement Leaves Verification to the Buyer

Purchase agreements usually narrow what a buyer may rely on. An anti-reliance provision is meant to establish that the buyer relied only on the representations written into the agreement, and not on the other statements and materials that moved during diligence. Whether a particular clause achieves that, and what it reaches, is a question for deal counsel. Delaware commentary marks an outer limit: courts will not let such a provision protect a seller from false representations made inside the contract itself.

The practical consequence belongs in the diligence plan. Anything a slide asserts that nobody converted into a written representation is, to the buyer, an unpriced belief. So the checking happens before signing, and it happens against the systems, because afterwards the agreement decides what counted.

Where the Evidence Behind Each Claim Lives

Each material claim has a home, and the home is a system somebody operates daily and nobody built for this deal.

• Revenue and cash. The general ledger, the bank statements and the invoice file. Tracing a presented figure back to an independent record is the oldest move in quality of earnings work, which adjusts for non-recurring items and normalizes revenue into a baseline a plan can stand on.
• Recurring revenue and churn. The billing or subscription system, which holds contract terms, start and end dates, and cancellations as entered rather than as summarized.
• Pipeline and win rates. The CRM, and specifically its stage history. Customer demand itself sits with commercial due diligence and its own outside sources.
• People. The HR information system and payroll, read together, for headcount, attrition and cost.
• Delivery. Source control, the issue tracker and the deployment history, which are the raw material of engineering due diligence.
• Service. The ticketing system, with response and resolution times per ticket rather than a tile that averages them.

Auditing standards rank evidence in a way that explains the list. Evidence from a knowledgeable source independent of the company beats internal evidence, and evidence obtained directly beats evidence handed over. A slide is internal and indirect. An extract pulled from a system whose operator has no stake in the framing sits higher. Operational due diligence is largely the discipline of climbing that ranking.

Ask for the Definition Before the Number

Request the written definition first, and read it before the figure. Public-company disclosure guidance is a useful borrowed standard here, although it binds nobody in a private deal: a presented metric wants a clear definition, a statement of how it is calculated, disclosure of any change in that calculation between periods, and enough surrounding context for a reader to understand what is shown. Ask a private company to meet that bar and the response itself is informative.

What is counted, what is excluded, and when the calculation last moved. Those questions do most of the work. Adjusted earnings deserve the same treatment: an adjustment that removes normal, recurring, cash operating costs the business cannot run without is not normalization, and a measure presented without an honest label and a plain description of what it excludes can mislead whoever reads it.

Request the Full Extract, Then Test It

A curated exhibit is not an extract. Ask for the whole population in the system's own export format, with the filter or query that produced it, and with the system fields intact. Then test the extract before using it. Auditing standards ask exactly this of information a company produces: test its accuracy and completeness, or test the controls over them, before relying on it.

• Completeness. Reconcile the extract totals to something outside the extract: the trial balance, the bank, the payroll register.
• Accuracy. Pull a sample of rows back to their underlying documents, the signed contract, the invoice, the offer letter.
• Boundaries. Confirm the period, the entity list and how conversions were handled, then rerun the slide's own calculation on the extract and see whether the slide reappears.

Only the last step is a comparison. The steps before it decide whether the comparison means anything.

Activity Records Show How a Result Was Produced

Totals answer what. Activity logs answer how. Most systems keep a chronological record of their own use: who changed a field, when a deal moved stage, when a credit was issued, when a deployment ran. Read those and the documented process meets the practiced one.

• Timestamp clustering. Stage changes that land in bursts at the end of a quarter describe a different sales motion from the one on the slide.
• Edits after the fact. Records amended once a period closed, or deals closed and reopened and closed again, change what a win rate means without changing what it says.
• Who touched it. When the same person appears on every material record, the result is real and the process is not repeatable. That is a leadership finding as much as a data one, and it belongs beside the Management Team Assessment.

How to Classify a Gap Between the Slide and the System

When the extract disagrees with the slide, the useful question is not which is right. It is what kind of disagreement this is, because the kind decides what changes. The sort below is mine, a way to file findings rather than a tested taxonomy.

• Definition. The slide metric is calculated differently from the system metric of the same name. Restate the series both ways and carry both forward.
• Population. The slide shows a subset, the extract shows everything. This one tends to move price, because excluded rows are rarely excluded at random.
• Timing. Cut-off, period boundaries, bookings read as billings. It bends the shape of growth more than its level, and it usually lands in the working capital and net debt mechanics.
• Behavior. The totals tie, and the activity records show the result was produced differently from the way it was described. Price may survive this. The plan after close will not, because whatever produced the number has to be rebuilt or retained on purpose.

A gap of any kind also says something about the people who prepared the slide. A management team that answers a definition request in a day, with a full extract and the query attached, is telling you how it runs the business. A team that answers with a fresh exhibit is telling you that too.

What a Finished Comparison Should Put in Front of the Deal Team

The output is not prose. It is a line for each material claim, readable across:

• The claim, quoted from the slide, with the slide it came from.
• The system of record holding its evidence, and who operates that system.
• The definition used, in writing, and who supplied it.
• The extract: scope, date, and which accuracy and completeness tests it passed or failed.
• The reconciliation, with the residual difference stated rather than smoothed.
• The kind of gap, and what it moves: price, terms, or the plan after close.
• What could not be tested, and why not.

That last line is the one that earns trust. This work is not an audit and it issues no opinion. Traced, tested and reconciled are the verbs. Proven is not. An operational diligence checklist decides which claims are worth chasing, and the comparison above is the part of due diligence that survives an investment committee.