Sector notes
Report the automation result in hours, not percent
A ratio with no baseline is not a result. When we restated a robotics line's outcome from an efficiency percentage to the time the same work now takes, the number became checkable — and, as it happens, larger.
Executive summary
24 h → 1 h
unboxing time for the same volume of stock on the Raymang Eggs line: by hand, and on the robotic line
3
things a result needs before anyone can check it: a baseline, a unit, and a window
Core conclusions
- A percentage gain with no stated baseline can describe almost anything, which is why it persuades nobody who is paying attention.
- The measured quantity — hours for the same work, units per shift, labour hours per batch — is smaller on the page and larger in effect, because it can be checked.
- Record the baseline before go-live. A result measured against a remembered baseline is an estimate wearing a decimal point.
For some time our site described the outcome of a robotic carton-handling line as an efficiency gain of over 1,000 percent. The figure was not invented. It was also not useful, and a reviewer said so: no baseline, no unit, no measurement window, on a page whose whole argument is that it does not do promotion. They were right.
The restated version is this. Unboxing the weekly import shipments by hand took the line 24 hours. On the robotic line, the same volume of stock is unboxed in one. That is the same fact the percentage was reaching for, and it works out to a larger ratio than the one we had been quoting. It is also a sentence a buyer can hold us to.
What a percentage hides
'Efficiency up 1,000 percent' could mean throughput, labour, cost, cycle time or floor space. It could be measured over a week or a year. It could be against the line's worst day or its average. Each of those gives a different number, and the ratio quietly picks one without saying which.
The problem is not that ratios are dishonest. It is that a ratio is a division, and a reader who is not shown the denominator cannot tell what was divided by what. On a plant floor that is the difference between a result and a claim.
The same result, stated two ways
| As it was stated | As a measurement |
|---|---|
| Efficiency gain of over 1,000 percent | 24 hours of unboxing, done in one, for the same volume of stock |
| Baseline: unstated | Baseline: the manual process, on the weekly import shipment |
| Unit: unstated | Unit: elapsed hours for a fixed quantity of work |
| A reader can be impressed | A reader can go and time it |
The second number is the one operations cares about
The hours figure is the headline, but the line changed a second quantity that never made it into the percentage. Imported eggs used to come up from cold storage to room temperature with industrial fans, four to five hours per batch, before anyone could handle them — a process that risked condensation and delayed every delivery. Chilled stock now holds at roughly 4°C and moves into a climate-controlled, low-humidity room to normalise only once it is scheduled to go out.
That is not an efficiency percentage. It is a food-safety and shelf-life change, and on a short-expiry product it is the one the operations manager would have led with. A ratio has no room for it. A measurement does.
How to record a result that will survive a reviewer
- Fix the baseline before go-live: the quantity, the unit, and the period it was measured over, signed off by the people who ran the old process. We have said this before about OEE, and it applies to every metric.
- Measure the new process on the same quantity and the same unit. If the unit has to change, say why.
- State the window: the period the after-measurement covers, and whether it includes commissioning. Where a window has not been recorded, say so rather than imply one.
- Report the measured pair. Let the reader do the division if they want the ratio; it will be the right one.
Twenty-four hours to one is a smaller number on the page than 1,000 percent. It is a much larger claim, because it is one we can be asked to prove.
