Eliminate Before You Automate: The Process Audit Most Firms Skip

A 58-person professional services firm mapped their project delivery workflow and found 16 steps. They were about to spend $40,000 automating all 16. Cendia asked one question about each step: has this step caught a real problem in the last 12 months? Nine of the 16 steps had not.

Eliminate-Before-Automate is a sequencing rule. It states that before any workflow is automated, every step in that workflow should be evaluated for elimination. The evaluation is simple: has this step caught a real problem — a genuine error, a compliance issue, a client-impacting failure — in the last 12 months? If the answer is no, the step exists out of habit, and automating it means paying to preserve a habit at machine speed.

Cendia’s finding across 20+ engagements: 40-60% of operational steps in services firms between 30 and 100 people can be eliminated before any automation conversation begins. The steps that survive elimination are the ones worth investing in — and the automation scope drops by half, which means the cost drops by half, the timeline compresses, and the failure rate plummets.

Why elimination comes first

Automation locks processes in place. Once a workflow is automated, changing it requires engineering time, testing, and migration — costs that didn’t exist when the process was manual. A manual process that includes unnecessary steps wastes staff time. An automated process that includes unnecessary steps wastes staff time, engineering time, and maintenance budget.

The Surface vs. Structure Lens reveals why this matters: the surface symptom is “this workflow takes too long.” The structural cause is usually that the workflow contains steps that no longer serve a purpose. Automating the workflow treats the symptom. Eliminating the unnecessary steps treats the structure.

A specific example makes the cost concrete. A 44-person accounting firm had a client deliverable review process with 11 steps:

Steps 3, 6, and 8 had either never caught a problem or hadn’t caught one in over a year. Step 3 alone consumed 45 minutes per deliverable across 15-20 deliverables per week — roughly 12-15 hours of associate time weekly on a review that had produced zero findings in 23 consecutive instances.

Eliminating those three steps saved 16 hours per week before any automation was discussed. At a loaded rate of $135/hour, that’s $112,320 per year in recovered capacity — from removing steps, not from buying technology.

The elimination test

One question applied to every step in a workflow: has this step caught a real problem in the last 12 months?

The question is deliberately specific. “Could this step catch a problem?” doesn’t count — any step could theoretically catch something. “Has this step ever caught a problem?” is too broad — a step that caught one issue 3 years ago and nothing since is running on insurance logic, not operational necessity.

Twelve months is the threshold because it balances risk tolerance with operational reality. A step that hasn’t caught a problem in 12 months of regular execution has had hundreds of opportunities to prove its value and hasn’t. The risk of removing it is low. The cost of keeping it is measurable.

Three outcomes are possible for each step:

Eliminate. The step hasn’t caught a real problem in 12 months. Remove it from the workflow. If anxiety about removal is high, implement a 90-day trial elimination — remove the step but keep the ability to revert. Cendia’s finding: fewer than 5% of eliminated steps are ever reinstated. The anxiety is almost always larger than the risk.

Document. The step has caught real problems but the decision criteria aren’t written down. Before this step can be automated, the judgment it contains needs to be explicit. Write the rule: what does the step check for, what constitutes a pass, and what triggers an escalation?

Automate. The step catches real problems, the decision criteria are documented, and the check can be performed by a system rather than a person. This is where automation investment belongs — on steps that are proven necessary and fully understood.

How the audit works in practice

The full Eliminate-Before-Automate audit takes one day for a single workflow. Here’s the protocol Cendia uses.

Morning (3 hours): Map and count. Pull the workflow’s current documentation. If none exists, map it from scratch by interviewing the 2-3 people closest to the work. List every step. Count them. Most workflows at services firms between 30 and 100 people have 10-20 steps. If you find fewer than 8, you’re mapping at too high a level.

Midday (2 hours): Apply the elimination test. For each step, ask the person who performs it: “When was the last time this step caught a real problem?” Record the answer. Be specific — “I think maybe a few months ago” needs a follow-up. “In March, we caught a client name error” counts. “I’m not sure it ever has” is an elimination candidate.

Afternoon (2 hours): Score and decide. Categorize each step as Eliminate, Document, or Automate. Calculate the time spent on elimination candidates — hours per week, multiplied by loaded rate, multiplied by 52 weeks. That number is your annual cost of unnecessary steps. Present it to leadership with the recommendation.

The Cost-Per-Workflow framework provides the financial context. Every eliminated step reduces coordination cost (the meetings and handoffs around that step disappear) and failure cost (the rework triggered when that step produces a false positive or delays the workflow). At firms in the 30-100 person range, eliminating 40% of steps in a high-volume workflow typically recovers $80,000-$200,000 per year in combined coordination and failure cost.

What firms get wrong about this audit

Two common mistakes derail the Eliminate-Before-Automate process.

Mistake 1: Treating the audit as a threat assessment. When teams hear “we’re eliminating steps,” they hear “we’re eliminating jobs.” The audit targets steps, not people. Recovered hours get redeployed to higher-value work. Frame the audit around capacity recovery, not headcount reduction.

Mistake 2: Keeping steps “just in case.” Risk aversion preserves unnecessary steps more than any other factor. A partner who remembers a formatting error from 2019 will fight to keep the format check step even though it hasn’t caught anything in 18 months. The 90-day trial elimination addresses this — remove the step with a documented revert plan. If a problem surfaces in 90 days, reinstate it. If it doesn’t, the step is confirmed unnecessary. Data resolves the argument that opinion cannot.

What happens after elimination

A workflow that started with 16 steps and dropped to 9 is fundamentally different to automate. Smaller scope, fewer edge cases, lower cost, and higher team confidence because every remaining step has proven its value.

The sequencing matters for the Three-to-One Rule as well. An automation project scoped at 16 steps costs $40,000 and saves an estimated 300 hours per year at $145/hour loaded rate — a return of $43,500, barely clearing the 1x mark and nowhere near the 3x threshold. The same project scoped at 9 steps after elimination costs $18,000 and saves an estimated 200 hours per year — a return of $29,000, clearing the 1.6x mark. Combined with the $112,000 already saved through elimination, the total return on the combined initiative exceeds 5x the total investment.

Without the elimination phase, the firm would have spent $40,000 to save $43,500. With it, they spent $20,000 to save $141,320. Elimination changed the math on everything that followed.

Nine of the 16 steps hadn't caught a real problem in over a year. The firm was about to pay $40,000 to automate all of them — preserving nine habits at machine speed.

What this isn’t

Scope notes:

• This isn’t an anti-automation position. Cendia recommends automation routinely — after the elimination test has been applied. The framework is about sequencing, not rejection. Automate what survives; eliminate what doesn’t.
• This isn’t zero-risk. Removing a workflow step carries the possibility that the step was catching something nobody noticed. The 90-day trial elimination manages this risk explicitly. In Cendia’s experience, fewer than 5% of eliminated steps are reinstated — but the revert option matters for stakeholder confidence.
• This isn’t applicable to regulated processes. Compliance-mandated steps — regulatory filings, audit trails, legally required reviews — exist because a regulation requires them, regardless of whether they’ve caught a problem recently. Apply the Three-Layer Compliance Model to those workflows instead. The elimination test is for operational steps, not regulatory ones.

FAQ

How do we decide which workflow to audit first?

Start with the highest-volume workflow that’s also being considered for automation investment. This gives the audit maximum financial impact: it reduces the automation scope (saving money) while recovering hours from eliminated steps (saving more money). Most firms choose client onboarding or project delivery as their first audit target.

What if the team disagrees about whether a step has caught a real problem?

Require specifics. “I think it’s caught things” doesn’t pass the test. “In April, this step caught a scope discrepancy that would have cost us $8,000 in rework” does. If nobody can cite a specific instance with a date and a consequence, the step is an elimination candidate. Specificity resolves disagreements that generalizations cannot.

Can we apply this to steps we’ve already automated?

Yes, and the savings can be substantial. Automated steps that no longer serve a purpose still consume maintenance time, integration complexity, and system resources. A quarterly review of automated workflows using the same 12-month test often identifies 2-3 automated steps that can be decommissioned — simplifying the system and reducing maintenance cost.

How does this relate to the 2-Minute Rule?

They’re complementary and sequential. Eliminate-Before-Automate is the strategic audit — it determines which steps in a complex workflow should exist at all. The 2-Minute Rule (Automation Triage) is the tactical starting point — it identifies the simplest tasks to automate first. Run Eliminate-Before-Automate on your complex workflows. Run the 2-Minute Rule on your recurring tasks. Together, they cover the full automation landscape.

This article is part of Cendia’s Operational Frameworks series. Companion pieces cover the 2-Minute Rule, the Handoff Cost Model, and Cost-Per-Workflow — the diagnostic toolkit Cendia uses to find where growing services firms can recover capacity before investing in technology.