Shanghai-listed manufacturer
Signal Collective · Case study
01 / 07
A 10-day engagement, a clear plan for the board.
A Shanghai-listed manufacturer had a rollout that had stalled, and a board that couldn't tell what was actually wrong. Over 10 days we mapped how the business ran, watched how the work really happened, and talked to every department. We reviewed 12 systems, found 14 fixes, and picked out the 6 that were worth the most. On day 10 we handed the board a clear plan of what to fix and what each fix was worth. The numbers below are exact.
10 days
12
14
6
Day 10
02 · The problem
02 / 07
The board knew something was off. Nobody could say what.
This is a listed manufacturer with several plants on the mainland. A push to modernise how the plants ran had stalled, and the reasons kept changing depending on who you asked. The board wanted a straight answer to three questions, and they wanted it fast. The big consulting firms quoted three to four months and a large bill. We offered a fixed 10-day project instead.
What's actually happening on the floor?
The board was reading reports, but the reports didn't match what people on the floor described. They wanted to know which one was real.
Where are we losing time and money?
Everyone had a theory. Nobody had put a number on it. The board needed to see the biggest leaks ranked, not argued about.
What do we fix first?
They didn't want another wish list. They wanted a short, ordered plan they could act on and hold people to.
03 · What we did
03 / 07
We mapped how the business really ran.
We didn't work from slides. We went on-site, followed the work from the floor to the boardroom, and looked at the 12 systems the business actually runs on — production, the warehouse, quality, energy, safety, HR, reporting, and the rest. For each one we asked a simple question: when two systems disagree, which one is right? In several cases nobody on the floor could tell us. That gap alone explained a lot.
01
Mapped the 12 systems
We laid out every system the business runs on, how data moves between them, and where one hands off to the next. We marked which system is the real record when two disagree.
02
Watched how the work happened
We walked the floor and followed a job end to end. What people did in practice often wasn't what the process said, and the gap was where things were breaking.
03
Talked to every department
Production, warehouse, quality, energy, safety, and reporting. We checked every finding with the people who owned it, so nothing in the plan was a guess.
04 · What we found
04 / 07
14 fixes. Here are five of them.
We found 14 things worth fixing. The full list stays with the client, but these five are the kind of thing we surfaced — each one confirmed with the department that owns it. They are common problems in a plant this size, so we can describe them without naming anything.
Two systems fought over the same number
Two production systems both claimed to own downtime data. Operators entered it in whichever was handy, and the number the board saw came from just one of them. A real chunk of downtime never showed up in the reports.
Reporting was stitched together by hand
Each quarter, one analyst pulled numbers out of several systems by hand and glued them together. It took more than two weeks, and two of the sources didn't stamp their data cleanly, so it never quite reconciled.
Handovers between shifts were lost
Shift notes were written on paper and scanned in after the shift ended. Problems that happened right at a shift change got closed out far less often than ones in the middle of a shift.
Energy use wasn't tied to output
Meters tracked total energy, but nothing linked it to a batch, a shift, or a product line. The plant could report how much energy it used, but not how much per unit — a number they were increasingly being asked for.
Supplier records were scattered
Proof of supplier quality lived in several places at once. Pulling a clean trail together — for an audit or a deal — meant assembling it by hand every time.
05 · What each fix was worth
05 / 07
Six fixes, each with a number the board could act on.
Out of the 14, we picked the 6 that were worth the most and put a number on each — money saved, revenue at stake, or risk removed. For every one we said who should own it and roughly how long it would take. Half could be done with the systems they already had; half needed some new tooling built.
Settle who owns the downtime number
Pick one system as the record and reconcile the other to it, so the board's numbers finally match the floor.
Automate the quarterly report
Wire the source systems together so the report builds itself instead of eating an analyst for two weeks each quarter.
Move shift handovers off paper
Run a digital handover on a couple of lines first, so problems at shift change stop slipping through.
Link energy to output
Connect the meters to batches and lines so the plant can report energy per unit, not just a total.
Put supplier records in one place
Move supplier quality evidence into a single record, so a clean trail is one click away instead of a manual hunt.
Set up how new tools get measured
Agree on how a pilot proves itself before it goes plant-wide, so good ideas don't stall in limbo again.
06 · The plan on day 10
06 / 07
What the board got, and what happened next.
On day 10 we sat down with the board and walked them through it in one sitting, in both English and Chinese. They got a clear picture of how the 12 systems fit together, the 14 fixes with the owner for each, and the 6 priority fixes with a number and a timeline. It moved the board from arguing about the problem to deciding what to do about it. Within two weeks, they asked us to help turn the top fixes into real, ready-to-build work.
A map of the 12 systems
How everything connects, and which system is the real record when two disagree. In plain language, not a data model.
The 14 fixes, with owners
Each fix tagged with how urgent it is, where it comes from, and who should own it. Nothing vague.
The 6 priority fixes, with numbers
The short, ordered plan the board asked for — what to do first, what it's worth, and how long it takes.
07 · Work with us
07 / 07
A note on this story
- The client's name is private. We don't share the company, its exact sector, its size, or its ticker.
- The numbers are exact. 12 systems reviewed, 14 fixes found, 6 with measured ROI, 10 days, plan on day 10 — these are the real figures from the engagement.
- The findings are described as patterns. The five examples above are the kind of thing we found, written generally. The client-specific detail stays confidential.