Ulfhednar Labs Article

Automation Should Remove Work, Not Move It

Many “automated” systems look efficient only because the labor moved out of the accounting line being measured. Real automation reduces total work across the process instead of exporting clerical burden to customers or other employees.

A branching workflow where repetitive tasks disappear into an automated processing core while a separate path leaves people with only the necessary decisions.
Efficiency should be measured across the whole workflow, not only inside the department buying the software.

A system is not more efficient when it saves one department ten minutes by making every customer perform fifteen minutes of unpaid clerical work.

The Spreadsheet Says It Was Automated

A company replaces a staffed process with a portal. The internal labor number falls. The project is declared a success.

Meanwhile, the customer now has to create an account, re-enter information the company already has, interpret internal terminology, upload documents in a specific order, check status manually, and call support when the automated workflow rejects something without explaining why.

The work did not disappear. It changed payroll.

Local Efficiency Can Create Global Waste

This is a classic systems problem. A team optimizes the metric inside its boundary without measuring the effect on the rest of the process. Ten minutes disappear from one department and reappear as fifteen minutes of customer effort, five minutes of support work, and another exception queue somewhere else.

The organization can still report “automation savings” because the transferred labor is not counted in the same place.

A better measure is total effort required to reach the outcome: employee time, customer time, exception handling, rework, failure demand, training, and the cost of correcting bad automated decisions.

Digitizing a Bad Process Preserves the Bad Process

Automation projects often begin by taking an existing paper or human workflow and reproducing each step in software. That can be useful as a transition, but it should not be confused with redesign.

OECD guidance on digital government has drawn the same distinction: digital transformation is not merely converting an existing process into electronic form; the deeper value comes from re-engineering processes and services around user needs. The technology is the enabler, not the reason the process exists.

If a five-step approval path exists only because three old systems could not share information, a new system should not automatically preserve the five steps.

Humans Should Keep the Work Humans Are Good At

Removing work does not mean removing people from every loop. Some tasks involve judgment, empathy, accountability, negotiation, creative interpretation, or handling the unusual case that the automated path cannot safely resolve.

Good automation absorbs repetition and prepares information so that human attention is spent where it has leverage. It can collect routine inputs, validate formats, reconcile records, route cases, generate drafts, surface anomalies, and make the state of the process visible.

The goal is not “no humans.” The goal is “no unnecessary human effort.”

Measure the Whole Journey

Before automating a workflow, map who does the work now. Then map who will do it afterward. Count the new accounts, clicks, handoffs, approvals, exception paths, support calls, and maintenance tasks—not only the labor line that becomes smaller.

The best automated system may look surprisingly simple to the user because the complexity has been absorbed where it belongs: inside the system.

Automation should remove friction from the process. When it merely pushes that friction downhill until someone else has to carry it, the system is not smarter. The accounting is.

Sources

Ulfhednar Labs

Software Without Limits.

Professional writing, development perspective, and ideas from the work behind Ulfhednar Labs.

More Articles ↗
All Articles