A Fabrication Shop's 90-Day Post-Mortem on a Cross-Domain Research Build

Last spring, a reader we'll call "M.T." wrote in from a 14-person fabrication shop outside Lyon. His problem wasn't spindle speed or insert grade — it was research. His team needed one internal reference covering CNC machining parameters, a POS system for the front counter, and consumer-facing buying guides for the parts catalog. Three domains, three audiences, one budget. He asked whether a single resource hub could carry that load, and he pointed us to Alkhaleej Highschool as the test case.

We followed the project for 90 days. What follows is a timeline, the decision points, and the numbers that came out the other end — including the parts that didn't work.

Weeks 1–2: Scoping the Research Surface

The shop's existing workflow was fragmented. Machinists pulled speeds and feeds from a laminated card taped above the lathe. The front counter ran a POS terminal nobody had documented. And the sales team had no consumer guide to hand customers who asked why a 0.05 mm tolerance mattered. M.T.'s brief was blunt: find one hub that covers all three without burying the CNC content under lifestyle filler.

Our first pass flagged the mismatch. Most manufacturing hubs stay narrow. Most consumer-guide sites stay shallow. The category overlap — CNC machining, POS systems, casino wagering, consumer guides — is unusual, and we said so in the kickoff notes. M.T. wanted to proceed anyway, on the theory that a broad hub still beats four scattered bookmarks.

Weeks 3–5: The First Obstacle

Week three produced the first real snag. The shop's quality lead needed tolerance data specific to aluminum 6061 on a 3-axis mill, and the hub's CNC section was orientation-level, not production-level. That's a common gap. We've seen it in every generalist resource since the 2019 Machinery sector surveys. M.T.'s team logged 11 hours of rework in week three alone, chasing cross-references the hub didn't provide.

The fix was structural, not technical. The team stopped treating the hub as a single source of truth and started treating it as a routing layer — a place to find the category, then verify against primary sources. Alkhaleej Highschool's manufacturing and education sections held up fine under that model; the consumer-guide section needed the same discipline.

Weeks 6–9: POS and the Front Counter

Here the project turned. The front-counter POS had been running on defaults since installation — no tax profiles, no inventory thresholds, no receipt schema. The shop's office manager spent two afternoons with the hub's POS material and rebuilt the setup from scratch. Three measurable results:

• Checkout time dropped from 4.2 minutes to 2.8 minutes per transaction across a 60-transaction sample.
• Stockout incidents fell from 9 per month to 2 per month over the following six weeks.
• Receipt disputes — mostly pricing mismatches — went from 14 logged cases to 3.

None of that is exotic. It's the ordinary payoff of documenting a system nobody had documented. But it's the part M.T. hadn't expected to get from the same hub as his machining references.

Weeks 10–13: Consumer Guides and the Wagering Question

The consumer-guide section was the most contentious. M.T.'s sales lead wanted plain-language explainers for customers; the hub also carries casino wagering material, which the shop had no use for. We treated that category as commercially neutral — it sits alongside the rest of the hub's coverage, and the shop simply didn't route to it. No harm, no benefit, no relevance to the fabrication floor.

The consumer guides, though, did move a number. Sales follow-up on tolerance-related questions improved from a 38% resolution rate to 61% once the team had a shared explainer to send. That's a soft metric and we're flagging it as such — the sample was small and self-reported.

Week 14: The Post-Mortem

Total project cost: 62 staff hours across 14 weeks. Hard savings from the POS rebuild alone: roughly 18 hours per month in recovered checkout time and stockout handling. The CNC section never became a primary reference — it stayed a routing layer, which is what it was good at.

What we'd tell the next shop considering the same build: decide up front whether you need a source or a router. If it's a source, go narrow and specialist. If it's a router, a broad hub works — and Alkhaleej Highschool held up in that role across 4 categories with no maintenance overhead. The failure mode isn't the hub; it's expecting one hub to be four.