November 13, 2026 yield analytics production quality

Production Yield Analytics: Measuring What Matters on the Line

Your CM sends a Friday afternoon email: "Batch 037 complete. 9,500 of 10,000 units passed. Yield: 95%. On track for shipment Monday." You read it, feel a knot in your stomach, and close the laptop. Ninety-five percent sounds fine. Most things in life would be great at 95%. But you just lost 500 boards.

Five hundred boards at $25 each is $12,500 in scrap. Add the rework labor your CM will charge for the ones that can be saved — maybe half, at $8 per board, another $2,000. Add the two days of shipment delay while rework runs. Add the 20 boards that pass rework but fail again in the field three months later because the root cause was never identified. Add the support tickets, the returns, the one-star reviews.

Ninety-five percent yield isn't a number to feel okay about. It's a signal that something on your line needs attention. Production yield analytics is how you find it.

The Number That Tells You Everything

First pass yield (FPY) is the percentage of units that pass every test the first time through, with no rework, no retest, no manual intervention (Deltek). It's the single most honest metric in manufacturing because it can't be gamed without changing the test program itself. Industry benchmarks put electronics PCB assembly FPY at 91–94% average, with top-quartile performers hitting 95–97% (OxMaint).

Your CM almost certainly tracks FPY internally. They know that Line 2 runs 97.8% on your product while Line 1 sits at 94.3%. They know the night shift drops half a point. They know that the new component lot from the alternate supplier has a slightly higher failure rate on the accelerometer calibration step. This data lives in their MES system, and it's extraordinarily valuable.

The problem is that you've never seen it. You get the Friday email: "95%." One number, no breakdown, no trend, no failure analysis. Not because the CM is hiding it — because there's never been a shared system to make it visible. Your CM's quality engineer has a Pareto chart on their desktop right now that would tell you exactly what's wrong. You just can't see it.

The Pareto Insight

Here's what that Pareto chart would show you: 80% of your failures come from two or three root causes. This isn't a guess — it's one of the most reliable patterns in manufacturing quality, known as the Pareto principle. Joseph Juran applied Vilfredo Pareto's 1906 observation to quality control, finding that most production defects arise from a small percentage of causes (Rework.com). In real factories, the split may be 70/30 or 85/15 depending on the process (Jodoo).

Imagine your 500 failed boards from batch 037. You pull the failure data and sort by test step. The microphone sensitivity test accounts for 210 failures. The BLE RSSI measurement catches another 145. The deep sleep current test flags 85. Everything else combined is 60. Three tests, three problems, 88% of your scrap.

Now you can ask the right questions. The microphone failures — is it a component issue? You check the component lot number. Every failing unit traces back to lot MIC-2026-1028 from your alternate supplier. The primary supplier's lot passes at 99.6%. The BLE failures — check the reflow profile logs. Oven 3 has been running 4°C above target since Tuesday when maintenance adjusted the conveyor speed. The deep sleep current — that's a firmware issue. Version 3.0.22 has a regression where the WiFi radio doesn't fully power down; the fix is already in 3.0.23 but hasn't been pushed to the line yet.

Three root causes. Three fixes. Switch back to the primary mic supplier. Recalibrate oven 3. Push firmware 3.0.23. Next batch runs 99.2%.

This is what production yield analytics does. It doesn't just tell you that something's wrong — it tells you what's wrong, and gives you enough context to fix it before the next batch starts.

Yield Drift: The Slow Problem

A sudden drop from 98% to 90% gets everyone's attention. What's harder to catch is the slow drift — 98.1% in week one, 97.4% in week two, 96.8% in week three, 95.9% in week four. Each week looks fine on its own. The Friday email says "96%, on track." But the trend is unmistakable once you plot it.

Yield drift almost always has a physical cause. A solder paste batch approaching its shelf life. Pogo pins on the test fixture wearing down after 40,000 cycles, increasing contact resistance and causing intermittent false failures (PCB&Assembly). A new operator on the SMT line who loads component reels slightly offset. A seasonal humidity change in the factory affecting solder joint quality.

Without trend data, you don't see drift until it becomes a crisis. With it, you catch the slope at week two and investigate. The CM's process engineer knows exactly what changed — they just didn't connect it to your yield numbers because nobody was watching the trend line together.

This is where shared analytics transform the relationship. When both you and the CM see the same yield curve in real time, nobody needs to send an email asking "what happened?" The conversation shifts from "your yield is dropping" to "let's look at what changed between week one and week two." That's the difference between blame and problem-solving.

Batch-to-Batch: The Pattern Finder

Individual batch yield tells you about that batch. Batch-to-batch comparison tells you about your process. When batch 037 runs 95% and batch 038 runs 99.1%, the question isn't "what went wrong with 037" — it's "what was different?"

The answers live in the metadata: component lots, operator shifts, equipment IDs, environmental conditions, firmware versions. A yield analytics dashboard that correlates test failures with these variables turns weeks of investigation into minutes. "Show all batches using component lot MIC-2026-1028" — there's your answer. "Compare yield by operator across the last 10 batches" — there's your training gap. "Overlay reflow oven temperature logs with BLE test failures" — there's your equipment issue.

Your CM already has most of this data. Their MES system logs operator IDs, machine assignments, and lot numbers as a matter of course. The missing piece is connecting it to your test results and making the correlation visible to both sides.

From Data to Dashboard

On the AN & Factory platform, yield analytics are automatic. Every test result that flows through the system — every pass, every fail, every measurement value — feeds into a dashboard that both the founder and the CM can see. First pass yield by batch, by day, by week. Failure Pareto by test step. Yield trend over time. Batch comparison with drill-down to component lots and equipment.

No extra work for the CM. The data is already being captured at the test station. The platform just makes it visible and correlatable. When yield drops, both sides see it at the same time. When a component lot causes elevated failures, the traceability chain identifies every affected unit instantly.

The CM's quality engineer — the one with the Pareto chart on their desktop — now has a shared screen. Their expertise becomes a joint asset instead of an internal report that never leaves the factory.

The 95% Problem, Solved

Go back to that Friday email. "Batch 037 complete. 9,500 of 10,000 passed. Yield: 95%."

With production yield analytics, that email doesn't exist. Instead, you open your dashboard on Tuesday and see yield trending down from 98% to 96.2% over the last 48 hours. The failure Pareto shows microphone sensitivity spiking. You click through — lot MIC-2026-1028, alternate supplier. You message the CM: "Let's hold the alternate mic lot and switch back to primary for the rest of this batch."

By Thursday, yield is back to 98.5%. The batch ships on time. The scrap pile is 150 boards instead of 500. You saved $8,750 and two days of delay. The CM is relieved because you caught it early enough to fix it mid-batch instead of after the damage was done.

That's what measuring the right things does. Not just counting passes and fails — but understanding why, watching for drift, and fixing problems while they're still small. Your CM's data is already rich enough to tell this story. All it needs is a shared screen.

Want to see your production yield in real time? AN & Factory provides automated yield dashboards, failure Pareto analysis, and batch comparison — all from your existing test data. Get in touch.

Sources

Have thoughts on this? Join the discussion on our forum.

← Back to blog