Key takeaways
- Every plant already collects shop floor data. The real cost is latency, transcription and reconciliation.
- Paper is instant to write and slow to use. Spreadsheets are legible but unjoined to the work order.
- A connected system removes reconciliation because the data is joined by design.
- Move one unanswerable question at a time — and retire the spreadsheet deliberately, or it never dies.
Every plant already does shop floor data collection. The question is never whether you collect it — it is how long the data takes to arrive, how much of it is guessed, and how many meetings are spent arguing about whose number is right.
There are broadly three approaches in use, often all three at once in the same building. It is worth being honest about what each actually costs.
Paper
Still extremely common, and not stupid. Paper is instant, needs no training, survives a power cut and never rejects an entry because a field is mandatory.
What it costs you is latency and transcription. The data exists the moment it is written and becomes usable somewhere between hours and weeks later, after someone types it in. Every transcription step introduces error, and the person typing is rarely the person who knows what the handwriting meant.
It also cannot alert anyone. A value drifting out of tolerance sits quietly on a clipboard until the sheet is collected.
Spreadsheets
The usual second step, and a genuine improvement: legible, sortable, and it does arithmetic. Most plants get a long way on spreadsheets, which is exactly why they are so hard to leave.
The costs arrive gradually. Versions multiply — the one on the shared drive, the one on the supervisor’s desktop, the one attached to an email in March. There is no reliable link between a row and the work order it describes, so joining production data to quality data means matching on text that humans typed. And the entry still happens after the fact, which means the floor is always reporting the past.
Watch out for
The spreadsheet that has become infrastructure. If month-end depends on one workbook that one person maintains, you do not have a reporting process — you have a single point of failure with a named owner. That risk rarely appears in any register until the owner is on holiday.
A connected system
The third approach captures data where and when it happens, against the record it belongs to: this work order, this operation, this machine, this material lot.
Two things change fundamentally. First, machines report what machines know — counts, durations, stoppages, process values — without anyone transcribing. Second, the data is already joined: quality results, material consumption and downtime sit against the same order, so no one has to reconcile them later.
The real cost of paper and spreadsheets is not the data entry. It is that every question requires a reconciliation before it can be answered.
An honest comparison
- Speed to usable data — Paper: hours to weeks. Spreadsheets: hours to days. System: immediate.
- Accuracy — Paper: transcription error. Spreadsheets: transcription plus formula drift. System: machine-read values are not transcribed at all.
- Joinability — Paper: manual. Spreadsheets: text matching. System: joined by design.
- Ability to alert — Paper: none. Spreadsheets: none in practice. System: real time.
- Cost to start — Paper: nothing. Spreadsheets: nothing. System: real, and the honest reason most plants delay.
That last row is the one worth taking seriously. The case for moving is not that paper is embarrassing; it is that at some point the reconciliation cost exceeds the implementation cost — and that crossover is different for every plant.
How to move without a big-bang project
The failed version of this is a plant-wide rollout that replaces everything at once, trains everyone in a week, and collapses back to spreadsheets within a quarter. A sequence that works better:
1. Pick one question you cannot currently answer. Material variance on one product. Downtime by cause on one line. Not “digitise the floor”.
2. Capture only what that question needs. Resist the temptation to collect everything because you are already there. Every extra field is a tax on every future entry.
3. Start with machines that already talk. Automatic capture requires no behaviour change and no compliance, which makes it the safest first win.
4. Prove the number, then show it back. When the floor sees data they entered turn into a decision within days, the next expansion is much easier.
5. Retire the spreadsheet deliberately. Running both indefinitely is the most common way these projects quietly die — people keep the workbook because it is familiar, and the new data is never trusted.
In practice
Keep the paper for a fortnight after go-live and compare the two records. It costs little, it finds the capture gaps quickly, and it is the fastest way to earn trust in the new numbers — which is what actually determines whether the change sticks.
Where Capitán fits
Capitán MES captures shop floor data against live work orders — reading machine values directly, holding quality control and waste alongside production, and carrying in-process traceability so the records are joined rather than reconciled. Because it is built on a no-code platform, you can start with one line and one question and extend without a new project each time.
If you are weighing the crossover point for your own plant, talk to our team.
The short version
Paper is instant but late. Spreadsheets are legible but unjoined. A connected system removes the reconciliation step entirely — and the way to get there is one unanswerable question at a time, not a plant-wide rollout.