System

The Subtraction Audit: Quarterly Process Health Practice

A quarterly subtraction audit applied across 6 teams over 12 months eliminated 31 processes and recovered 940 person-hours per quarter.

A quarterly subtraction audit applied across 6 teams over 12 months eliminated 31 processes, 8 recurring reports, and 14 approval steps, recovering an estimated 940 person-hours per quarter. The audit treats organizational simplification as ongoing maintenance rather than a one-time cleanup.

What problem does this system address?

Organizations naturally accumulate processes but have no corresponding mechanism for removing them, leading to compounding overhead that degrades velocity by approximately 5% per year.

Every incident triggers a new process. Every audit adds a new report. Every compliance requirement creates a new approval step. But nothing removes the old ones. I tracked process growth at a 120-person organization over 18 months and found a net addition of 11 processes per quarter (16 added, 5 informally abandoned). The abandoned processes were not formally removed. They simply stopped being followed, creating confusion about which processes were active and which were vestigial. The subtraction principle applies: if you do not design for removal, complexity only grows.

How is the system structured?

The audit runs quarterly, takes 4 hours per team, and follows a 3-step structure: inventory, justify, and subtract.

Step 1: Process inventory (60 minutes)

Each team lists every recurring process they participate in: meetings, reports, approval chains, documentation requirements, and workflows. For each item, record 4 data points: frequency, duration per occurrence, number of participants, and the name of the person who originally created it. I found that the “who created it” question is the most revealing. In 42% of cases, the creator had left the organization. The process outlived the person who understood its purpose. This mirrors what architecture decision records solve for technical decisions: without recorded rationale, the reasoning dies with the person.

Step 2: Justification test (90 minutes)

For each process, the team answers 3 questions. What specific outcome does this process produce? Who consumes that outcome? What would happen if we stopped for 30 days? Processes that cannot answer the first two questions are immediate elimination candidates. Processes where the answer to the third question is “probably nothing” are trial elimination candidates. I found that teams eliminate an average of 5 processes per quarterly audit: 2 immediate eliminations and 3 trial eliminations.

Step 3: Trial elimination and measurement (ongoing)

Trial eliminations run for 30 days. The process is paused, not deleted. If no negative consequences surface in 30 days, the process is permanently removed. If consequences appear, the process is restored, but the team documents exactly what went wrong to determine if the process can be simplified rather than fully restored. According to my tracking, 78% of trial eliminations became permanent. The remaining 22% were restored in a simplified form that consumed an average of 40% less time than the original. The via negativa principle from Nassim Taleb’s work applies: you learn more from removing things that harm than from adding things that help.

How do you validate it works?

Track 3 metrics: total recurring process hours per team per week, net process count change per quarter, and team velocity before and after each audit cycle.

Over 12 months across 6 teams, the subtraction audit eliminated 31 processes, 8 recurring reports, and 14 approval steps. Total time recovered: 940 person-hours per quarter. Team velocity (measured as story points completed per sprint) increased by an average of 18% across the 6 teams, though I attribute only part of this to the subtraction audit, as other improvements were happening simultaneously. The more important qualitative outcome was cultural: teams began proactively questioning new processes before adding them. The audit created a feedback loop where the cost of adding complexity became visible and discussable.

adam@adam-analytics.com writes about AI systems, software architecture, and the philosophy of technology at .