B04 · CNC
RUNNING
More output, from the same shift.
A line can be full of working machines and still run below its pace. The loss isn't in any one machine. It is in the waiting, the backtracking and the one fiddly step nobody timed. We measure it, cost it, and hand you the case for fixing it.
1 · pick base4s
2 · seat clips6s
3 · fit lid clip5–30s
4 · press tabs5s
5 · inspect4s
6 · refill binidle
step 3 variability 76% · invisible on a walkaround
on this order
30,000
units assembled by hand
the case it made
Redesign the part
costed, evidenced, funded
replace
Manual logging
where reliable automatic evidence is practical
manual work
measured
per step, per unit, per shift
improvement projects
evidenced
quantified before and after, then watched
The cells that report nothing are setting the pace.
A press streams its own state second by second, with nobody lifting a finger. The same work done by hand produces no signal at all. It is usually the hand work that decides how fast the line can go.
01 · No signal to read
The automated cells report themselves. The manual ones sit blank.
So the improvement effort goes where the data is, not where the constraint is.
B07 · press
RUNNING
Bench 4 · hand
NO DATA
Pack · hand
NO DATA
02 · The measurement tax
Every way of closing that gap costs you the thing you're trying to win back.
A clipboard, or a tablet and a form. Asking an operator to time their own work steals the very seconds you are chasing. The figures come back late and half remembered. So it never sticks.
×Stop, note the time, key it in, then carry on
×Three seconds a unit is twenty-five hours across one order
×Figures rounded to the nearest guess, trusted by no one
There is a third option: measure it without asking anyone to.
3s logged per unit 30,000 units
Time spent logging 25 hours not assembling
An OEE for the cell that never had one.
A camera watched bench 4 for a shift. Nobody stopped, nobody keyed in a thing. Those timings become the three numbers you already track, so the manual bench sits in the same view as the CNC cell, comparable like for like.
61%
OEE this shift
first number this cell has ever had
availability
88%
performance
72%
quality
96%
largest loss: step 3 variability · worth chasing first
On one line assembling 30,000 storage boxes at a time, vision caught a single fiddly step that usually took
seconds, but every few units blew out to thirty.
Invisible on a walkaround. Real money across a full order. A clear case to redesign the part.
Measured, comparable, improvable: the manual work joins everything else Insights already sees, and stops being the part of the floor you have to take on faith.
Three ways we lift the pace.
Manual assembly is often a black hole: no timings, no view of variance, and finished stock that lands in one unplannable jump. Computer vision times each step without interrupting the operator, and counts output automatically, so planning sees product move in real time.
Before
No timings, no view of variance, and finished stock that appears in one unplannable jump.
With DataQI
Every step timed and output counted without interrupting the operator, so variance becomes a costed case.
Vision tracks flow across the floor, showing where product queues, where operators backtrack, where time quietly leaks between steps, turning a vague sense that it feels slow into a specific place to fix.
Before
A line runs below pace for reasons nobody can pin down, hidden in how work moves between steps.
With DataQI
Flow is measured across the floor, so queues and bottlenecks become visible and can be designed out.
Improvement effort tends to follow whoever is loudest, and proving a finished project worked is hard. So genuine wins go unfunded and regressions go unnoticed. Insights ranks opportunities by evidenced impact, quantifies each project before and after, and keeps watching so a gain that slips away is caught early.
Before
Effort goes where attention is loudest, and proving a finished project worked is hard, so wins go unfunded.
With DataQI
Opportunities ranked by evidenced impact, projects quantified before and after, regression caught early.
Outcome 04
Keep the knowledge →
Thirty years of know-how, one question away, before it retires.
The hard partComputer vision →
How the manual work gets measured, on your hardware, on your terms.
StartDiscuss a production problem →
Pick one manual process. We'll show you the data it's been producing all along.