Deal money does not vanish at random. It disappears in the same five places, deal after deal, and every one of them is a data failure wearing a financial costume.
I started thinking of these as crime scenes after a conversation with an M&A accounting specialist. The framing stuck because it is accurate. Each one has a body, a weapon, and a motive. The body is enterprise value. The weapon is a number nobody could trust. The motive is almost always that the data was never built to survive scrutiny.
If you want to know where deals lose money, you do not need a forensic accountant for the first pass. You need to know which five rooms to walk into. Here they are.
Crime scene one: the financial story
The first place money disappears is the story the financials tell, or fail to tell.
A buyer is not paying for last year’s EBITDA. They are paying for a number they believe will repeat and grow. Everything in the financial narrative exists to make that belief defensible. Revenue by segment. Margin by product line. The shape of the recurring base. Customer concentration. The bridge from reported numbers to adjusted numbers.
When the underlying data cannot support that story cleanly, the story gets discounted. Not because the business is bad, but because the buyer cannot tell the difference between a strong business and a well-told one. Ambiguity always prices in the buyer’s favor.
I have sat with sellers who genuinely could not produce a consistent revenue-by-segment number across two different reports. Sales had one cut. Finance had another. The board deck had a third. None of them were wrong on purpose. They were built at different times by different people against different definitions. But to a buyer, three versions of revenue is not a reporting quirk. It is a reason to assume the worst version is the true one.
The data discipline that prevents this is unglamorous. One definition per metric that finance, sales, and operations all sign. One source the number comes from. The goal is not perfect data. It is a number nobody in the room relitigates. When the financial story rests on numbers that hold up under questioning, the discount disappears.
Crime scene two: working capital
The second crime scene is the one that pays out, or costs you, on the day of close: the working capital adjustment.
Most deals set a working capital peg, a normal level of working capital the business is expected to deliver at close. Land above it and the seller gets paid more. Land below it and the price comes down dollar for dollar. It is one of the few places in a deal where a data error converts straight into cash with no negotiation buffer.
The problem is that working capital is built from the messiest data in the company. Accounts receivable aging that nobody has scrubbed. Inventory valued on assumptions that drifted years ago. Accrued liabilities that live in spreadsheets. Prepaids and deferred revenue that depend entirely on how a few judgment calls were made.
When that data is soft, the peg becomes a fight. The buyer’s team rebuilds the historical average from the ground up, finds the soft spots, and every soft spot moves the peg in their direction. A receivables book that has not been written down to reality. Inventory that includes stock that will never sell. Each adjustment is small. Together they move real money on close day.
The discipline here is to treat working capital as a data exercise long before it is a negotiation. Scrub the receivables. Validate the inventory. Document the accruals so they survive a hostile rebuild. The seller who walks in with a defensible, well-documented working capital history controls the peg. The seller who improvises hands the buyer a free lever.
Crime scene three: diligence
The third crime scene is diligence itself, and specifically the quality of earnings review.
This is the room where the buyer’s accountants take the financial story apart and decide how much of it to believe. Every adjustment they cannot get comfortable with comes out of the number. A quality of earnings gap is not a moral judgment about the business. It is the gap between what you reported and what you can prove.
I have written before about how data failures create these gaps in the causes of QoE gaps, and about how the cumulative effect can cost a real fraction of the multiple in how data problems cost one company half a turn. The pattern is consistent. The business is fine. The data describing the business is not, and the adjustments pile up.
The vivid version of this is the night before closing. I have seen a deal where, hours before signing, the diligence team surfaced a pool of work that had been delivered but never billed. Real revenue, genuinely earned, sitting in a state where nobody could prove when it would convert to cash or whether it ever would. It had simply fallen between two teams’ systems and never been invoiced.
That single discovery threatened to blow up the number. Not because the work was fake, but because nobody could produce clean data to show its status. The seller knew the revenue was good. The buyer saw an unbilled balance with no audit trail and started subtracting. A number that should have been a footnote became a midnight negotiation, because the data behind it was never built to be looked at.
The discipline is to run your own diligence before the buyer runs theirs. Find the unbilled work, the soft adjustments, and the numbers that cannot be proven while you still have time to fix them quietly. Diligence is not a test you pass by being honest. It is a test you pass by being provable.
Crime scene four: earnouts
The fourth crime scene shows up after close, sometimes years after: the earnout.
Earnouts exist to bridge a disagreement about value. The buyer will not pay today for performance they have not seen, so part of the price is tied to the business hitting targets after the deal. On paper it is elegant. In practice it is one of the most reliable sources of post-close litigation, and the reason is almost always data.
An earnout is a bet settled by a number. Hit the revenue or EBITDA threshold and the seller gets paid. Miss it and they do not. But the two sides frequently cannot agree on what the number even is, because after close the business changes. The buyer reclassifies revenue. Allocates new costs. Integrates systems that redefine how a metric is calculated. Suddenly the figure that determines a large payment depends on accounting choices the seller cannot see and did not agree to.
Earnout disputes are rarely about bad faith. They are about ambiguity. If the agreement does not pin down exactly how the earnout metric is calculated, on what data, with what adjustments excluded, then both sides are free to compute a number that favors them. And they will.
The discipline is precision, written down before the deal closes. Define the earnout metric to the level of the data source and the calculation method. Specify which post-close changes are excluded from the math. The cleaner the data definition at signing, the smaller the room for a fight at settlement. Vague earnout language is a dispute scheduled for a future date.
Crime scene five: integration
The fifth crime scene is the longest and the quietest: integration.
This is where the value creation plan is supposed to pay off, and where data failure does the most cumulative damage. The deal thesis assumed cross-sell, or consolidated procurement, or a single view of the customer, or faster reporting across the combined entity. All of it depends on data from two companies coming together into something usable.
When it does not, the plan stalls. I have watched a unified customer view scoped for ninety days take more than a year because three CRMs would not reconcile and nobody could agree on which customer record was the real one. Every month that initiative sat paused was a month of value the model assumed and the business did not deliver. That is integration value loss, and it is the form of M&A value leakage that rarely makes the board deck because no single report shows it.
The damage is hard to see precisely because it is spread across time. There is no line item for “the cross-sell that never launched” or “the procurement savings that arrived eighteen months late.” The money does not disappear in one event. It leaks, quarter after quarter, while the integration that was supposed to take a season takes years.
The discipline is to treat integration data readiness as part of the deal, not an afterthought for the operating team to discover. Map every initiative in the plan to the data it requires. Know before close whether that data exists, whether it can be trusted, and whether it can actually be joined across the two businesses. The initiatives that depend on data nobody has built are the ones to question first, because they were never deliverable on the timeline the model assumed.
The common thread
Walk all five crime scenes and the same culprit keeps turning up. The financial story, the working capital peg, the diligence review, the earnout, the integration plan. Different rooms, different dollar figures, one weapon.
It is not that the businesses are weak. It is that the data describing them was never built to withstand a buyer who is paid to find the cracks. Ambiguity does not split the difference. It prices against you, every time.
The fix is the same discipline in all five rooms, applied early. One definition per metric. One source per number. A documented trail that survives a hostile rebuild. None of it is glamorous, and none of it shows up as a headline accomplishment. But it is the difference between a clean exit and a midnight negotiation over money that was yours all along.
I wrote about how this leakage continues quietly after close in the hidden half-turn. If you want to see how a buyer scores your data before they put a number on it, walk through the Buyer Scorecard and find your own cracks before they do.