Case study: three spreadsheets of quality data, one plastics processor

A mid-sized custom molder came to us last year with a familiar problem. Quality data existed. It just lived in three places: a shared Excel workbook the QC tech updated at end of shift, a second workbook the process engineer kept for capability studies, and a third one the plant manager used to roll numbers up for the Monday meeting. The three didn't agree. They never had.

They weren't looking for a platform. They wanted the Monday number to match the shift number.

What they had

Twelve presses, mixed tonnage, running a wide product mix for automotive and consumer customers. A historian on the newer cells, nothing on the older ones. A CMMS used mostly for PM scheduling. First-pass yield was tracked by part number but reconciled by hand against scrap tickets a day or two later. Dimensional checks were written on a paper log at the press, keyed into the QC workbook later, and sometimes transcribed a third time into a customer PPAP packet.

This is not an unusual setup. Baytech's writeup on reshored manufacturing describes the same pattern across the mid-market: critical operations running on spreadsheets and disconnected tools, with the integration happening in someone's head.

What we built first

One thing. A floor-side capture app running on ruggedized tablets at each press. The operator records the dimensional check and the scrap reason at the station, against the job and the shot count, with a dropdown for cause codes the quality manager actually uses. Behind it, a small service that normalizes the entries and writes to a single database the three spreadsheets now read from, instead of being the source of truth.

We did not replace the historian. We did not touch the CMMS. We did not propose an MES. ProcessApp's paper traveler case study makes a point we've seen hold up: the biggest change from capture-at-the-station isn't the dashboard, it's that the data stops drifting between the floor and the office.

What changed

The Monday number matches the shift number. Scrap reasons are specific enough that the process engineer can actually sort them. The QC tech got about an hour of her shift back because she isn't retyping the paper log. Nobody has quoted us a yield improvement because they're still establishing a clean baseline, which is the honest answer.

Why it kept going

Once the data was in one place, other questions became answerable. Changeover timing by tool. Scrap correlated to resin lot. A simple weekly email to the plant manager flagging presses whose cycle time had drifted more than a set percentage from the setup sheet. None of this was in the original scope. Each piece was a week or two of work on top of a foundation that already existed.

That's usually how the long engagements start. Not with a platform decision. With one small thing that works, and a reason to build the next one.