[ CASE / 05 ] Manufacturing more out the door

A mid-sized manufacturer running a single facility with a few hundred line workers.

The floor and the finance side finally agree. One number, not three spreadsheets.

  • Factory floor
  • Live rollup
  • Quality alerts
  • Always on
[ 05_01 ]

The issue

how it came to us

Their factory-floor system and their finance system ran completely separately. The production schedule lived in one, the financial reality in the other, and the only bridge between them was an analyst emailing a summary spreadsheet every Friday morning. Plant managers were making Tuesday's decisions on the previous Friday's data. Quality problems got caught at final inspection — sometimes after the next batch had already started. Their operations chief said the gap between what was actually happening on the floor and what the system claimed was the single biggest drag on their margin.

[ 05_02 ]

Discovery & analysis

how we figured it out
  1. A month of floor activity, lined up against the finance side, surfaced a set of warning signs that reliably showed up before a quality problem downstream.
  2. Two weeks on-site filled in the rest — walking the line with each shift supervisor, cataloguing every workaround: the laminated cheat sheets, the side chat groups, the foreman with his own spreadsheet of what the system gets wrong.
  3. How they measured output didn't survive the review either; with the operations chief we agreed it was quietly undercounting the small stoppages — exactly where the lost output was hiding.
[ 05_03 ]

How we worked with the team

no over-the-fence handoffs
  • We had a short daily stand-up with the plant manager for the first six weeks — the unwritten rules he carried in his head became the alerting logic.
  • One of our engineers spent a week on the night shift, where most of the quality issues actually happened, to see the failure modes for real.
  • The line supervisors had their own channel for the alert rollout and tuned the sensitivity themselves.
[ 05_04 ]

What we built & shipped

the system in operation
  • The floor activity now feeds a rollup every ten minutes instead of once a week, and it writes back to the finance system and to a single shared dashboard.
  • When one of the warning signs crosses the line, an alert goes to the right supervisor with the batch, the machine, and a link to that line's recent history.
  • An hourly summary using the corrected measure (small stoppages included) goes to the team. The plant manager's view shows the same number as everyone else's — no more arguing over whose spreadsheet is right.
[ 05_05 ]

Outcome

measured, not modeled
Cost saving

Catching quality problems on the line instead of at final inspection saves a substantial six-figure sum a year, plus far fewer rushed shipments to recover from missed deliveries.

Speed

Output up by roughly a fifth. Visibility went from weekly to hourly. Acting on a line problem went from several hours to about ten minutes.

Accuracy

Measuring output properly exposed a chunk of capacity that was hiding in small stoppages — they recovered most of it within a couple of quarters.

  • The "what the system gets wrong" cheat sheet has been retired — by the foreman who used to keep it, willingly.
[ 05_06 ]

Where it stands now

we don’t hand off

It's been running a good while now, and we haven't handed it off. Early engineering at the moment: using the same warning signs to predict when a machine needs servicing before it breaks.

Got a process like this?
Talk to a founder.