Chapter 1
A quote, a full schedule, and a $280,000 answer
Ridgeline's plant manager is quoting a 240,000-board-a-year job — about 40,000 six-up panels through SMT. The scheduler says the lines are full: SMT-1's board shows it loaded 92% of available hours. So there's a capex request on the table for a new placement machine at $280,000 — with a used one at $112,000 plus $40,000 install and programming as the fallback being argued about in the hallway. Before anyone signs either version, one question worth asking:
Not utilization. Not "the line was scheduled." Actual good boards, out the far end, as a share of the time the line was supposed to be making them. Hold that number — the plant's own data is about to answer it.
Chapter 2
No sensors. No MES. No subscription.
Here's the part the automation vendors don't want said out loud: Ridgeline already generates every number an OEE baseline needs. Thirty days, three sources they already have, plus one piece of paper. Tap each one:
Chapter 3
Three C-plus numbers make an F
OEE is availability × performance × quality. Each of SMT-1's three numbers, alone, looks survivable. Watch what multiplication does to them:
Availability
Of 480 scheduled hours, the line was actually running 340.8.
Performance
While running, it produced at 82% of the machine's real ideal rate.
Quality
93% of boards passed first time — straight off the audit sheets.
OEE
Where 480 hours went
Every bar is computed from the sources above — and they reconcile: good boards ÷ ideal rate = the last bar, to the decimal. Definition in the open: planned time here is all scheduled production hours, and every stop — including planned changeover time and meetings — counts against availability (the Six Big Losses convention). Define it differently and the number moves; that's the Week-1 conversation of the sprint.
The whole fleet, measured the same month
Left column: what the scheduler's board says (loading). Right: what the machines say (OEE). The cranky 1998 wave line everyone complains about? Second-best on the floor. Its failures are loud, so they get fixed — SMT-1's losses were quiet.
Fleet average: 58% OEE — close to the industry-typical 60. World-class is 85 (leanproduction.com/oee). Nobody at Ridgeline is bad at their job. They just couldn't see the number.
Chapter 4 · Playable
139.2 lost hours, sorted
Availability was the biggest of the three losses, so here's SMT-1's downtime, from the paper tally, sorted largest first. Click the bars — the biggest one opens.
Drill-in: 22 changeovers, planned vs. actual
Feeder changeovers are planned at 2.0 hours. The dots are the 22 that actually happened this month — average 3.4, and not one made plan. One convention, in the open: every changeover minute counts as availability loss here, planned or not. File it under "planned downtime" and the number stops hurting — and stops improving. And the boards built in the first 45 minutes after each changeover failed at six times the rate of the rest of the run — 284 of the month's 1,174 rejects in those windows alone. The changeover bar isn't one problem. It's the plant's biggest improvement project, pre-scoped.
Nobody gets blamed. That's the point.
Look at the reason codes: not one of them is a person. Changeovers run long because the feeder carts are staged in the wrong order. The padded cycle standard was set by a careful engineer being conservative in 2019, and nobody ever reset it. Reason codes indict the process, not the people — the second shift lead put it best: "Finally. Proof it's the feeders, not us."
Chapter 5 · Playable
What a point of OEE is worth
Ridgeline's fleet baseline is 58%. Drag the target. The hours are computed from planned time; the dollar figure uses a machine-hour value you can change to yours.
Formula in the open: 6 lines × 5,760 planned hrs/yr (the 480-hour month you just watched, twelve times) × points recovered ÷ 100, then × your machine-hour value. That's capacity you can go get — output on machines you already own, run the way you already run. It is not a savings claim, and not every shop finds a machine hiding in the losses. The honest promise is smaller: know your real number in 30 days.
A dashboard doesn't make parts.
True. So the baseline month at Ridgeline didn't end with a dashboard — it ended with three projects off the Pareto: SMED on the changeover bar (3.4 hrs → 2.2) plus startup halved, which together took availability 71 → 77; restaged feeder carts that killed the micro-stops, lifting performance 82 → 89; and the dial-in program plus the audit-driven defect fixes from Case File 01 taking quality 93 → 96. Ninety days later SMT-1 measured 77 × 89 × 96 = 65.8 — call it 66% OEE. Twelve points, no new machine: about 41,000 six-up panels a year of recovered capacity, and the 240,000-board job needed 40,000. It fit on the fleet they already owned.
Chapter 6
What it looks like on a Tuesday
No MES, no sensors bolted to anything, no subscription. The operator entry is a Microsoft List — five minutes a shift, replacing the paper tally. The pipeline runs in Ridgeline's own tenant. And at 7:00 every morning, this is on the quality office wall:
Honest cost on your side: the tally discipline (2 minutes a shift on paper during the baseline, 5 in the List after), and a 30-minute weekly review while the definitions settle. If your second shift pencil-whips the tally, the baseline says so — variance between shifts is itself a finding, and we'll show it to you rather than smooth it.
Chapter 7
The OEE Baseline Sprint
Thirty days, fixed price, and you own everything at the end — zero ongoing fees.
Baseline Sprint — up to 4 machines
Week 1 on-site: define OEE together, on your floor — what counts as planned downtime, where ideal cycle comes from — and design the tally sheet. Weeks 2–4: capture runs while we build the pipeline in your tenant (Lists/Forms → Power Automate → Power BI). Day 30: a baseline per machine, the loss Pareto, your top-3 losses dollarized, and the live board left running.
Baseline Sprint — up to 10 machines
Same sprint, whole floor. Same deliverables, same 30 days, same ownership: it's your tenant, your data, your dashboard when we leave.
The follow-on menu, prices in the open
Every bar on your Pareto is a candidate project. Each is a fixed-scope sprint, quoted up front, and each one is optional — the baseline stands on its own:
The questions worth asking us first
"My guys will game the tally."
Maybe — for about a week. Then the numbers argue with the machine counter and the audit sheets, and the tally loses. That's why the baseline uses three sources that have to reconcile, not one that has to be trusted.
"Is this surveillance?"
The tally has nine reason codes and none of them is a name. OEE measures the system: the staging, the standards, the scheduling. The fastest way to kill the program is to use it on people — so we won't, and we'll tell your leads that in the kickoff, out loud.
"We looked at MES. It was $200k."
MES is a fine year-three purchase. This is the $3,500 version of the question you actually have right now: what do my machines really run, and what's the biggest loss? If the baseline says you need more instrumentation, you'll buy it knowing exactly why.
"Our data's a mess. It'll take you months."
The baseline needs a cycle counter, box counts, reject counts, and one sheet of paper. If your audit data is the mess — that's literally Case File 01, and fixing it is how Ridgeline got the quality number in this story.
What does your "about 80%" actually run?
Thirty minutes, no deck. Bring the machine you're thinking about buying more of. We'll walk the OEE definition against how you actually run, and tell you what a 30-day baseline would and wouldn't answer — before you spend a dollar.
Book the 30 minutesIf it looks like a fit, the next step is the $3,500 baseline — 30 days, fixed, and you keep everything. If a baseline wouldn't tell you anything you don't already know, we'll say so.
Or just email john@bluestarcentral.com. New here? Start with Case File 01: The Daily Audit Problem or Case File 02: The Tribal Knowledge Problem.