Work I led before founding Concentric Lab.

From 20 hours to 20 minutes.

Every acquisition target required 20 hours of manual pre-diligence before the team could determine whether it warranted a formal evaluation. I built an automated screening system that produces a qualified advance-or-pass decision in 20 minutes.

Situation

The former employer's acquisition pipeline evaluated 50 to 80 acquisition targets per quarter. Each target requires the same pre-screening workflow: download three years of tax returns, identify owner add-backs, normalize compensation, calculate Seller's Discretionary Earnings, flag structural risks, and produce a one-page screening memo. The output is binary: advance to formal diligence, or pass.

The manual version took approximately 20 hours per target. At 50 targets per quarter, that committed 1,000 hours of analyst time to data extraction before a single strategic conversation occurred. Worse, approximately 70 percent of screened targets were eliminated in the first pass. The majority of the investment was wasted on deals that should have been filtered in the first hour.

The company was also paying $15,000 per formal Quality of Earnings report for deals that sometimes failed the pre-screen criteria. The QoE should be the final validation step, not a substitute for structured pre-screening. The cost structure was backwards.

Approach

I applied the screening framework. The pre-diligence workflow was the same every time: same document types, same calculations, same red-flag patterns, same output format. That is not analysis. That is a function waiting to be written.

The screening system I built ingests tax returns and financial statements in their original format. A document classification layer identifies the document type and tax year. An extraction pipeline pulls the relevant line items into a normalized schema. The calculation engine applies standard add-back identification, owner compensation normalization, and SDE computation.

The risk scoring layer evaluates twelve structural signals: revenue concentration, customer dependency, owner involvement ratio, margin trajectory, working capital patterns, and seven others calibrated against the company's historical deal data. Each signal produces a scored assessment, not a binary flag. A target with 40 percent customer concentration in a fragmented market scores differently than the same concentration in a market with three buyers.

The output is a one-page screening memo: calculated SDE, identified add-backs with supporting line items, risk scores across all twelve dimensions, and a recommended action (advance, conditional advance, or pass) with the reasoning chain visible. The analyst reviews the memo, not the raw data. The system did the extraction. The human makes the judgment.

Build time was eight weeks. I used the company's financial data from previous acquisitions as the training corpus for the add-back identification model and risk scoring calibration.

Outcome

Screening time dropped from 20 hours to 20 minutes per target. The bottleneck shifted from data extraction to document upload. Once the financials are in the system, the screening memo generates in under three minutes. The remaining time is analyst review and judgment.

Deal flow capacity increased 10x without additional analyst headcount. The team could screen targets it previously would have skipped due to time constraints. The volume increase surfaced two acquisition opportunities in the first quarter that would not have been evaluated under the manual process.

The pass rate at pre-screening remained consistent at approximately 70 percent, which validated that the system's screening criteria aligned with human judgment. The difference: the 70 percent elimination now costs 20 minutes per target instead of 20 hours.

QoE spending dropped because formal diligence is now reserved for targets that have already passed a structured pre-screen. Every dollar spent on QoE is spent on a deal worth evaluating.

20 min

Screening time (was 20 hrs)

10x

Deal flow capacity

8 wk

Build duration

12

Risk signals scored

What compounded

The screening system began as an internal tool at a former employer. I led this work before founding Concentric Lab. The extraction, risk scoring, and human-review pattern informed my later work.

This was not a Concentric Lab client engagement or acquisition program.

Related reading

Start a conversation.

Tell me which work keeps landing on your team's plate. I'll tell you which agent I'd build first and what it would cost.

Or email hello@concentriclab.com directly.

Prefer to talk? Book a 20-minute call.

You'll hear back from me within one business day. No sales team. No discovery call sequence.