Small tasks create large operating costs
Copy an enquiry into a CRM. Send a standard reminder. Rename and file a document. Check whether a payment arrived. Move a record to the next stage. Recreate the same report. Notify the same people when the same event happens.
None of these tasks looks important enough to justify a project on its own. That is why they survive. But commercial impact is a function of time multiplied by frequency. A small task repeated across employees, customers and working days can consume a surprising amount of payroll.
Consider a Cape Town property management company. Each tenant query arrives by email, is copied into a spreadsheet, forwarded to the right portfolio manager and logged again when it closes. Each step takes a minute or two. Nobody would describe it as a burden. Yet across several administrators and a few hundred queries a month, the company is paying for a meaningful slice of somebody's job purely to move information from one place to another.
Why businesses tolerate repetitive work
Manual processes often begin when transaction volumes are small. A competent employee creates a workaround and the workaround becomes the process. Growth then increases frequency while the method remains unchanged.
Microsoft's 2025 Work Trend Index describes a capacity gap in which leaders want higher productivity while employees report insufficient time or energy. Microsoft also describes coordination and administrative work as a tax on the work people were hired to perform. Its 2026 workflow guidance makes the operating problem explicit: when systems do not connect, people become the bridge and manually move work between platforms.
There are also human reasons. Repetitive work is predictable and feels productive. It is rarely measured, so it never appears in a management report. And the people who do it are often too busy doing it to redesign it. The result is the pattern described in why revenue growth requires more admin staff.
Warning signs
- Staff describe part of their job as "updating", "capturing" or "checking" information that already exists elsewhere.
- The same report is rebuilt by hand every week or month.
- Requests for extra administrative headcount track transaction volume closely.
- Errors cluster around copying, re-keying and filing rather than around decisions.
- Work stops when one person is on leave because only they know the routine.
How manual repetition becomes money
- Direct payroll: employees are paid to execute predictable steps.
- Opportunity cost: skilled people spend less time selling, serving, analysing or improving.
- Error and rework: repetitive copying and checking creates avoidable mistakes.
- Delay: work waits for a person even when the required decision is deterministic.
- Scaling cost: higher volume forces the business to add administrative capacity.
- Management overhead: supervisors chase completion because the workflow has no built-in state or exception handling.
The calculation most businesses never perform
Annual manual labour cost = task frequency per day × minutes per task ÷ 60 × loaded hourly labour cost × working days
If a five-minute task happens 80 times a day, it consumes 400 minutes, or 6.67 hours, every working day. At a loaded labour cost of R220 per hour and 240 working days, that is roughly R352,000 of annual labour capacity. The example is illustrative; your frequency and labour cost determine the real figure.
Then ask what proportion automation can genuinely remove. Do not assume 100%. Good automation normally leaves exceptions, judgement and oversight with people. If, in this illustration, a realistic design removes the routine 70% of instances and people still handle the rest plus monitoring, the recoverable capacity is closer to R246,000 a year, before the cost of building and maintaining the automation. The detailed method is set out in how to calculate the real cost of a repetitive task.
How to gather honest numbers
Estimates from memory tend to be wrong in both directions. Better evidence comes from the systems themselves: email volumes into a shared mailbox, records created per day, invoices processed per month. For time per task, sit with the person doing the work and time ten real instances, including the interruptions and look-ups. Record how many instances were exceptions. That exception rate is often the single most important number, because it determines how much an automation can actually take over. Much of this work is invisible in normal reporting, which is the subject of how to find the manual work nobody is measuring.
| What to record | Where to find it |
|---|---|
| Frequency | System logs, mailbox counts, record creation dates |
| Minutes per instance | Timed observation of real work |
| Exception rate | Count of instances needing judgement or a look-up |
| Error rate | Corrections, credit notes, customer queries |
| Loaded labour cost | Payroll plus benefits and overheads, from finance |
The commercial signal: automate stable, high-frequency, rules-based work where the current labour cost is measurable and exception handling can be designed safely.
Automating the wrong process can make things worse
Automation is not automatically good. If a process contains unnecessary approvals, duplicated data, unclear rules or dozens of exceptions, automation can simply make the bad design harder to change. The first question is therefore not "what tool should we buy?" It is "which steps should exist at all?"
That distinction matters because low-quality automation can create a second operating burden: somebody must now maintain the automation while employees continue to work around the underlying process. A common version is an automation that handles the straightforward cases while staff quietly keep handling everything else by hand, described in manual work persists because automation covers the happy path.
IBM's overview of business process automation makes similar points from a practical angle: good candidates tend to be high-volume, repetitive and time-sensitive, processes should be well documented before they are automated, and organisations new to automation do better starting small and measuring as they go.
The trade-off is between speed and durability. A quick automation built around today's workaround delivers savings sooner but locks in the workaround. Redesigning first takes longer but produces a process that is cheaper to automate and easier to maintain.
When this becomes commercially urgent
Priority rises when staff overtime increases, teams request additional headcount for administrative volume, customers experience delays, the same errors recur, or managers spend time policing routine completion. Another strong trigger is when transaction volume cannot grow without adding people in roughly the same proportion.
What good looks like
A mature automated workflow has a clear trigger, documented business rules, accountable ownership, logging, exception handling and measurable outcomes. The automation removes low-value execution while preserving human judgement where it matters. The business measures hours saved, errors reduced, cycle time improved or capacity created. Freed capacity is deliberately redirected to work that needs people, rather than quietly absorbed.
What to do first
Pick one team and one month. Ask them to list every task they repeat at least weekly, then measure the three most frequent using real system counts and timed observation. Calculate the annual cost of each, remove any step that adds nothing, and only then decide whether automation, integration or a simpler process is the right answer.
Practical next actions
- Ask each team to list repetitive tasks performed daily or weekly.
- Measure frequency and minutes, not just annoyance.
- Calculate annual labour capacity consumed.
- Remove unnecessary steps before automating anything.
- Prioritise high-frequency, low-judgement processes first.
- Design exception handling and monitoring before deployment.
- Measure realised savings after automation rather than relying on projected ROI.
Bottom line
Repetitive work is expensive because it hides inside normal payroll. The commercial case for automation appears when you stop evaluating each task individually and calculate the annual cost of repetition across the whole process.
Sources and further reading
- What Is Business Process Automation? - IBM (last verified 10 October 2026)
- Maximize Productivity with AI Automation - Microsoft (last verified 10 October 2026)
- https://www.microsoft.com/en-us/worklab/ai-at-work-intelligence-on-tap-will-reshape-knowledge-work (last verified 10 October 2026)
- https://www.microsoft.com/en-us/worklab/work-trend-index/2025-the-year-the-frontier-firm-is-born (last verified 10 October 2026)
Related Mellorca reading
- Gathering Information Before Work Can Start: The Payroll Cost of Preparation
- Manual Automation Triggers Leave Half the Process Manual
- The Cost of Automations That Stop Silently