Picture the first morning of a quality of earnings (QoE) engagement on a lower-middle-market acquisition. The seller has sent over what they describe as “everything you need:” a QuickBooks export, a folder of bank statement PDFs across three accounts, a few Excel schedules, and a pair of lease agreements. The buyer’s advisor needs to get to work analyzing the business.

Instead, the first two days are consumed by a task that produces no analytical insight whatsoever: getting the data into a format the team can actually use.

This is not an edge case. It is the standard experience for every deal team running merger and acquisition due diligence on a privately held business in the middle market ($5 million to $50 million enterprise value range). Before a single EBITDA adjustment can be evaluated, someone on the team has to import the books, map the chart of accounts, reconcile the bank statements, and normalize the financial statements.

Our research puts the average time spent on this data preparation work at eight to 15 hours per engagement—and that number rises sharply when the books are inconsistent, the formats are non-standardized, or the seller’s controller has been creative with account names.

The analytical work that follows—the work that requires judgment, experience, and professional skepticism—is not what takes the time. The data plumbing is.

Why AI Can Now Handle This

Artificial intelligence has been applied to financial data problems for years, but the technology had not yet evolved to meet the reliability threshold required for due diligence work—where errors have real consequences—until recently.

The combination of large language models (LLMs) that understand document context and purpose-built training on financial data patterns has changed the equation.

The core challenge in lower-middle-market QoE is that no two sets of books look alike. A chart of accounts at a manufacturing company uses different labels than one at a service business, and neither matches what an audited company would produce. Bank statement formats vary by institution and change over time. General ledger exports from QuickBooks, Sage, and NetSuite each have their own structure and quirks. For years, this variability defeated automated approaches. Today, however, AI models trained on large enough and diverse enough financial document sets can handle it reliably.

The result is that an analyst who previously spent the first two days of an engagement building the financial model from scratch now arrives at that same starting point in a matter of hours.

It’s important to note that time is not saved by cutting corners. Reconciliation and account mapping still happen, but the output is simply produced faster, with exceptions surfaced for human review rather than discovered through exhaustive manual comparison.

A New Entry Point: The Flash Proof of Cash

One of the more consequential applications of this technology is what we at QoEAgent AI call the Flash Proof of Cash—a rapid preliminary cash proof designed to be completed in the early stages of due diligence, before a full engagement is formally scoped.

In lower-middle-market transactions, the period between a letter of intent and the start of full diligence is often where deals die quietly. A buyer’s team has limited information, limited time, and a seller who may not yet be fully cooperative with data requests.

The Flash Proof of Cash is designed for exactly this window: a high-confidence preliminary reconciliation that surfaces material discrepancies or confirms that the books hold together quickly enough to inform the go/no-go decision before significant diligence resources are committed.

This is not a replacement for a full proof-of-cash engagement. It is a triage tool that lets advisors and their clients make better-informed decisions about where to focus time and attention once full diligence begins.

For deal teams managing multiple active engagements, the ability to quickly screen financial integrity before committing to a full scope has meaningful implications for both efficiency and risk management.

What This Means for the Lower Middle Market

The lower middle market is where the QoE data problem is most acute, and where AI-assisted automation offers the greatest relative benefit.

In the middle and upper markets, sellers typically have audited financial statements, cleaner books, and finance teams experienced with the due diligence process. In the lower middle market, none of those conditions can be assumed. The books are messier, the formats are more varied, and the data preparation burden falls entirely on the buyer’s advisor.

Reducing that burden does not change what good QoE work requires. It changes how quickly and efficiently that work can get started. The experienced transaction accountant who reviews a proof-of-cash is using the same professional judgment they have always used. But with AI integrated into the process, they are now able to spend that judgment on the findings rather than on the extraction.

For ACG members who advise on, finance, or close lower-middle-market transactions, the practical question is not whether AI will change the due diligence process. The question is how quickly the tools available to your deal teams will reflect that shift.

 Josh Cashman, CFA is an adjunct professor of Finance at University of Denver Reiman School of Finance and co-founder of QoEAgent AI, a Denver-based financial technology company that licenses AI research from the University of Colorado Boulder to automate quality of earnings analysis for M&A due diligence. He is a Chartered Financial Analyst (CFA) and M&A advisor with experience across lower-middle-market transactions, and a member of ACG Denver.

QoEAgent AI licenses AI research from the University of Colorado Boulder, developed under Dr. Tom Yeh, who now serves as our chief technology officer, to automate the document understanding and data classification work that has always sat at the bottom of the QoE engagement stack. The platform handles four specific tasks: proof-of-cash reconciliation, trial balance mapping, GAAP-structured financial statements, and contract and agreement extraction. Collectively, these account for the majority of data preparation time in a typical engagement. QoEAgent AI is available for early access engagements at qoeagent.ai.

ACG Insights is produced by the Association for Corporate Growth. To learn more about the organization and how to become a member, visit www.acg.org.