A company spent six months automating their invoice approval process. The automation worked perfectly—routing invoices through seven approval layers in seconds instead of days. Yet nobody was happy. Invoices still took forever to pay, finance was drowning in exceptions, and vendors were complaining more than ever.
The problem: they automated a broken process. The seven-layer approval chain was bureaucratic nonsense that served no purpose. Automation made the nonsense faster, but it was still nonsense.
This is the automation trap: assuming that removing manual steps equals process improvement. It doesn't. Sometimes you just end up with a faster way to do the wrong thing.
The Difference Between Automation and Improvement
Automation removes human effort from a process. Process improvement makes the process better. These aren't the same thing.
Consider expense report approvals. The manual process: Employee submits report. Manager reviews. Finance reviews. CFO reviews for amounts over $500. Accounting processes payment.
Automation option 1: Automate the existing workflow. Route reports through the same four steps automatically. Reduce approval time from 5 days to 2 hours.
Automation option 2: Redesign then automate. Question whether you need four approval layers. Discover that 90% of reports are routine expenses under $100. Create two paths: Auto-approve routine expenses under $100. Route only unusual or high-value expenses for human review.
Option 1 is faster. Option 2 is better. Option 1 automates waste. Option 2 eliminates it.
Why Organizations Fall Into The Trap
The automation trap is seductive for several reasons:
It's easier. Automating the existing process is straightforward—just replicate what people do now. Redesigning the process requires thinking, challenging assumptions, and navigating politics.
It's faster to implement. You can start coding immediately. Process redesign requires analysis, stakeholder interviews, and consensus building.
It feels safer. You're not changing how things work, just who does them. Process changes threaten people, challenge expertise, and create resistance.
It shows quick wins. Automation delivers measurable time savings immediately. Process improvement benefits are harder to quantify and take longer to materialize.
But quick, easy, and safe often means wrong.
The Cost of Automating Broken Processes
Automating bad processes creates three problems:
1. You Lock In The Problems
Manual processes are flexible. When something doesn't make sense, people work around it. Automated processes are rigid. Once the workflow is codified in software, changing it becomes a development project.
A manufacturing company automated their quality control process. The manual process had an unnecessary step where inspectors logged readings in two different systems because "we've always done it that way." Automation replicated this—now software logs data in two places. Fixing it requires changing code, testing, and deploying. The inefficiency is now permanent infrastructure.
2. You Scale The Dysfunction
Automation multiplies whatever you automate. If the process is efficient, you scale efficiency. If the process is wasteful, you scale waste at machine speed.
An HR team automated their interview scheduling process. The manual process was convoluted: Recruiter emails candidate with three options. Candidate picks one. Recruiter checks interviewer availability. Interviewer not available. Recruiter emails candidate with new options. Repeat.
Automation made this faster but didn't fix it. Now the system could handle 10x more scheduling confusion in parallel. Candidates got response emails faster, but still went through the same back-and-forth. The process was terrible. Automation made it terribly efficient.
3. You Miss The Root Cause
Many processes are inefficient because of poor design choices made years ago under different constraints. Automating without understanding why the process exists means missing opportunities for fundamental improvement.
A financial services firm automated their client onboarding process, replicating a workflow that required 47 different data points. Investigation revealed: most of those fields were added over time for regulatory requirements that no longer applied. Half the data was never used. Automation perpetuated a process designed for a regulatory environment from a decade ago.
How to Avoid The Trap: Redesign First
The solution is simple but not easy: fix the process before automating it. Here's how:
Step 1: Map The Current Process Honestly
Document what actually happens, not what's supposed to happen. Shadow people doing the work. Note every step, every handoff, every approval, every exception.
Ask: Why does each step exist? What would break if we removed it? How often do exceptions occur? What causes them?
Step 2: Identify The Waste
Look for these common sources of inefficiency:
Unnecessary approvals: Multiple sign-offs that add no value or catch no errors.
Redundant data entry: The same information entered in multiple systems.
Artificial handoffs: Work passed between people/teams that could be done by one person.
Defensive steps: Actions taken only to cover someone's liability, not to add value.
Historical artifacts: Steps that made sense five years ago but no longer serve any purpose.
Step 3: Redesign The Process
Before writing any code, ask:
What's the actual goal? Not "approve expense reports" but "ensure spending is legitimate and properly recorded."
What's the minimum viable process? What's the simplest workflow that achieves the goal?
Can we eliminate steps? Do we need four approval layers or would one suffice?
Can we handle exceptions differently? Can we auto-approve 80% and manually review the 20% that's unusual?
Can we reduce data requirements? Do we really need 47 fields or would 12 suffice?
Step 4: Then Automate
Only after redesigning should you automate. Automate the improved process, not the broken one.
A Framework: The Three Questions
Before automating any process, answer three questions:
1. Is this process actually necessary?
Sometimes the best automation is elimination. Does the process exist because it adds value or because it's always existed?
2. Are we doing this the right way?
Assume the process is necessary. Is the current workflow the best way to accomplish it? What would you design if starting from scratch?
3. What should be automated vs. what should be eliminated?
Not every step deserves automation. Some steps should be removed entirely. Some steps need human judgment and shouldn't be automated. Focus automation on steps that are genuinely repetitive, high-volume, and well-defined.
Real-World Example: Redesign Then Automate
A healthcare provider wanted to automate patient intake forms. The manual process required patients to fill out six pages of forms, much of which duplicated information already in their system.
Option 1 (The Trap): Digitize the six-page form. Let patients fill it out on a tablet instead of paper. Same 200 fields, just digital.
Option 2 (Redesign First): Question why six pages exist. Discover that half the questions are for billing, half for clinical care, and many are redundant. Redesign: Pre-fill everything already in the system. Ask only new information. Split into two focused forms—one for billing, one for clinical updates. Reduce from 200 fields to 30.
Result: Patient intake time dropped from 25 minutes to 7 minutes. Error rates decreased because patients weren't exhausted and rushing through redundant questions. Staff loved it because they got clean, relevant data instead of pages of noise.
Automation was part of the solution, but process redesign delivered most of the value.
The Bottom Line
Automation is powerful but not magical. It multiplies whatever you feed it—efficiency or waste, clarity or confusion, good processes or broken ones.
Before automating, ask whether the process deserves to exist at all. Before coding, redesign. Before scaling, fix.
The goal isn't to make bad processes faster. The goal is to eliminate bad processes and automate the good ones. Redesign first. Then automate.
