Automating an inefficient process does not fix it. It locks in the inefficiency, makes it harder to change, and adds a layer of capital cost on top. Lean value stream mapping before an automation project is the single most effective way to ensure your investment delivers what the proposal promised.
A food manufacturer in regional Victoria invested $165,000 in a cobot cell to automate tray loading at the end of their packaging line. Throughput did not improve. The robot was faster than the two operators it replaced, but the downstream palletiser was already the constraint. They had automated around the problem, not into it. The lesson cost them six months and a significant portion of their automation budget.
The Waste Amplification Problem
Automation does not fix process problems. It freezes them, makes them harder to change, and attaches a capital cost to them. A manual process that runs at 65 percent OEE for identifiable, solvable reasons will run at 65 percent OEE after you automate it, with one important difference: the automation equipment now limits your ability to cheaply modify the process when you understand the fix.
This is the central argument for lean analysis before automation investment. The Toyota Production System principle of jidoka (build in quality, stop and fix problems rather than passing them downstream) applies with equal force to capital investment decisions. Understand the process thoroughly before you commit to a technical solution.
Value Stream Mapping: What It Is and How to Do It for an Automation Decision
A value stream map (VSM) is a visual representation of every step a product takes through your facility, from raw material receipt to finished goods despatch, including the information flows that trigger each step. The current-state VSM captures what actually happens, not what the process sheet says should happen.
For an automation decision, the critical data points on each process step are: cycle time (the time to complete one unit of work at that station), changeover time, uptime percentage, and the inventory buffer before and after the step. These four numbers, mapped across every step in the value stream, reveal the constraint and the real sources of production loss.
Calculating takt time
Takt time is the rate at which you need to produce to meet customer demand. It is calculated as net available production time divided by customer demand rate in the same period. If a line runs 7.5 hours of productive time per shift (450 minutes) and customer demand requires 900 units per shift, takt time is 30 seconds per unit. Any process step with a cycle time longer than 30 seconds is a constraint. Any step with a cycle time significantly shorter than 30 seconds has spare capacity that is being wasted.
Identifying the Real Constraint Before Specifying a Solution
Eliyahu Goldratt's theory of constraints states that every system has exactly one constraint that limits throughput, and that improving any step that is not the constraint produces no improvement in system output. This is not a theoretical observation. It is a description of physics. Improving the speed of a non-constraint step only builds more inventory in front of the actual constraint.
The constraint in a production system is almost always identifiable by observation: it is the station with the most work-in-progress accumulating in front of it, the most frequently elevated heart rate among operators, and the longest average cycle time relative to takt. If you are not certain where the constraint is before you start writing an automation specification, you are not ready to write an automation specification.
The most common automation mistake
Specifying an automation solution for a process step that is not the system constraint. The automation delivers a faster cycle time at that step. Overall line throughput does not change. The project fails to deliver its business case. We see this pattern regularly in proposals that were written by vendors who were shown one station, not the whole value stream.
The Eight Wastes in an Automation Context
The TIMWOODS framework identifies eight categories of waste in manufacturing processes. Each has specific implications for automation investment.
| Waste | What it looks like | Automation implication |
|---|---|---|
| Transport | Moving product between stations over long distances | Conveyors solve transport waste, but only if the layout is already rational. Automating irrational transport embeds it permanently. |
| Inventory | Large buffers between process steps | High WIP buffers are a symptom of process instability. Automating a process that produces highly variable output does not reduce the inventory needed to buffer that variability. |
| Motion | Operator reaching, bending, walking within a station | Ergonomic redesign before automation often delivers 20 to 40% cycle time reduction with zero capital cost. |
| Waiting | Operator or machine idle time waiting for the previous step | Waiting waste is constraint evidence. Fix the constraint before automating the step that is waiting. |
| Overproduction | Producing more than the next step needs right now | Automated systems can produce overproduction far faster than manual ones. Ensure demand-pull signals are part of the automation design. |
| Over-processing | More work than the customer requires | Automating an over-processing step does it faster but still adds no value. Eliminate the step first if possible. |
| Defects | Rework, scrap, customer returns | Automating a process with a high defect rate without first addressing root causes automates defect production at higher speed. |
| Skills (underutilisation) | Experienced operators doing tasks that add no value | Automation should free skilled people for higher-value work. If the automation design still requires the same operator count doing low-value monitoring, the opportunity is not fully captured. |
SMED Before Capital: The Changeover Opportunity
Single-Minute Exchange of Die (SMED), developed by Shigeo Shingo at Toyota, is a methodology for reducing machine changeover time by separating internal setup activities (done with the machine stopped) from external activities (done while the machine is running) and converting as many internal activities to external as possible.
On a typical Australian packaging line, changeover times of 60 to 120 minutes are common. Applied systematically, SMED consistently delivers 40 to 60 percent reductions in changeover time with minimal capital investment, primarily through standardised tooling, pre-kitting of changeover parts, and clearly documented changeover procedures. Before investing in automated recipe management or quick-change tooling systems, a SMED workshop on the current process should be your first step. The residual changeover time after SMED gives you the accurate baseline against which automated changeover systems should be evaluated.
The Right Sequence: Lean First, Automate Second
The practical sequence for an automation investment that reliably delivers its business case is: map the current state value stream, eliminate or reduce all identifiable waste that can be addressed without capital, re-measure the process after lean improvements, then specify automation to hold the gains and improve further from a stable, understood baseline.
This sequence takes longer than going straight to an automation specification. It typically delays a project by six to twelve weeks. In our experience, it also consistently produces automation investments that deliver their projected ROI, rather than investments that achieve technical success but miss their production targets because the process they were designed for was not the process that existed.