Manufacturing · Field notes

Small manufacturing: throughput, rework, and the KPI wall

Tim Harmantzis · ApolloAI · ~8 min read
TL;DR

Throughput is a story about queues

In a 15-person shop, the constraint moves daily: a late raw receipt, a single skilled setup, a QC hold, a customer change. Big plants hide slack; small plants feel every hour. The winning habit is visibility—what job is here, why, since when—without turning the floor into paperwork theater.

Rework without blame games

Most rework clusters around a few defect types. Name them, count them, fix the top one with a mistake-proof or checklist at the source—not a lecture. People don’t enjoy doing jobs twice; systems make double work inevitable when specs travel by word of mouth.

The KPI wall that fits on one sheet

You can drown in metrics. For many SMB plants, on-time ship, first-pass yield (or simple scrap/rework rate), and order-to-cash cycle tell 80% of the story. Everything else is a drill-down.

Engineering and ops share one customer: the person who pays the invoice. When drawings, travelers, and as-builts disagree, the machine runs anyway—and finance hears about it later.

This week: plant reality check

  1. Walk the floor with a stopwatch mindset—where is static inventory (WIP piles)?
  2. List last month’s top five rework reasons in plain English; pick one for a 30-day experiment.
  3. Traveler audit: three random jobs—did the floor packet match what sales promised?
  4. Daily 10-minute stand-up: stuck jobs only— who unsticks, by when.
  5. Supplier subplot: late materials as percent of delayed ships—name names.

Where lean meets practical automation

Automation helps when it removes re-keying: quotes to work orders, BOM sync, simple alerts when a gate slips. It hurts when it adds fields nobody reads. Fit the tool to the standard work, not the other way around.

Want a short, ruthless process sprint on your constraint?

Book a free discovery call →

← All field notes