The data existed. It just lived in eleven different spreadsheets that nobody cross-referenced until month-end — by which point the problem had already cost real money.
This case study illustrates a composite scenario based on patterns observed across multiple client engagements. Names, figures, and specific details have been adapted to protect client confidentiality.
A ₹55 Cr precision tooling manufacturer was losing a consistent, frustrating amount to scrap and rework every month — somewhere between 4% and 6% of production cost, depending on the month. The number moved around enough that no one could say definitively whether it was getting better or worse, or why.
The plant wasn't short on data. Production logs, quality inspection sheets, machine downtime records, and material batch tracking all existed — across eleven separate spreadsheets maintained by five different people, none of which talked to each other. By the time the monthly quality report consolidated all of it, any pattern worth catching was three to four weeks old.
The MD's request wasn't about AI. It was simpler: "I want to know about a problem before it shows up in next month's numbers." Getting there required rebuilding how the business related to its own data, not buying a new tool.
A structured review against the Data Maturity Ladder — Untracked, Recorded, Reviewed, Predictive — placed the business at a deceptive midpoint: technically "Recorded," but functioning closer to "Untracked," because nothing was reviewed on a cadence that mattered.
Production output, defect rates, machine downtime, supplier batch quality, and rework hours were each tracked in separate spreadsheets, maintained by different people, on different update schedules. No one — including the MD — could pull a single, current view of plant performance without manually reconciling all five, a process that took a full day each time it was attempted.
Defect occurrences were logged daily at the point of inspection, but the data sat untouched until the monthly quality meeting. A defect spike that started on day 3 of a production run wasn't discussed until day 30 or later — by which time several more weeks of the same defect had already occurred.
Five different people generated data. No one was responsible for synthesizing it into a single view or flagging anomalies. The MD assumed the quality manager was watching for early warning signs; the quality manager assumed that was a function of the monthly report itself. The gap belonged to no one.
Once consolidated, the data revealed something none of the five individual spreadsheets had shown on their own: defect rates spiked consistently within 48 hours of receiving material from one specific supplier — a pattern invisible when production and supplier data lived in separate files, but obvious within minutes once they sat side by side.
Built one lightweight dashboard pulling production output, defect rate, downtime, and supplier batch data into a single view — refreshed daily, not reconciled manually once a month. No new software platform; the existing data sources were simply connected and visualized.
Result: What previously took a full day to reconcile manually is now visible in real time, automatically.Assigned the quality manager explicit, formal ownership of a twice-weekly anomaly review — 20 minutes, fixed time, fixed checklist — rather than leaving "watching the data" as an implicit, unowned responsibility.
Result: Time to detect a quality drift dropped from roughly 3 weeks to under 2 days.Once the supplier-linked defect pattern surfaced, the plant implemented an incoming batch inspection specifically for that supplier's material — catching defect-prone batches before they entered production, rather than after.
Result: Scrap and rework cost fell 22% within the first full quarter after the dashboard went live.| Position on Data Maturity Ladder | → | Untracked → Reviewed |
| Time to detect a quality drift | → | ~3 weeks → ~2 days |
| Scrap and rework cost | → | −22% |
None of this required new software spend, new headcount, or an AI platform. It required consolidating data that already existed, assigning ownership of watching it, and building a habit of reviewing it on a cadence that mattered — twice a week, not once a month.
The supplier-linked defect pattern had likely existed for over a year. It simply never had the chance to be seen, because the data that would have revealed it lived in two different spreadsheets that nobody had ever placed side by side.
Most SMEs aren't short on data. They're short on a habit of looking at it, and a single place to look.
Before investing in AI or new reporting platforms, the higher-leverage question is simpler: does the data your business already generates get reviewed on a cadence that could actually catch a problem early — and does someone specifically own that review? In this case, moving from "Recorded" to genuinely "Reviewed" on the Data Maturity Ladder did more than any new technology would have.
The Readiness Assessment includes a full Data & AI Enablement pillar — the same questions that surfaced this manufacturer's path from Untracked to Reviewed.