![[AN & Factory]](/static/img/logo.svg){kind=logo}
Why Your CM Isn't the Problem — Your Infrastructure Is
The forum post reads like a hundred you've seen before: "We're getting 4% field failure on our BLE sensor. Third batch in a row. Is it time to switch CMs?"
The replies are unanimous. "Fire them." "We switched to a factory in Dongguan and our yield doubled." "You get what you pay for." The founder reads these, nods, spends three months qualifying a new manufacturer, transfers tooling, runs first articles, and starts production. Two batches later, the field failure rate is 3.8%. Different factory, different city, nearly identical problem.
This story plays out constantly in hardware. And almost every time, the founder was asking the wrong question.
The factory you can't see
Here's what most founders don't realize about their contract manufacturer: the CM almost certainly has better quality data than the founder does. Walk onto a typical CM's production floor and you'll find MES dashboards tracking every stage, operator logs documenting rework events, AOI stations catching solder defects at rates above 99% (Predictable Designs), and test stations generating measurements on every board that passes through.
The production manager in Weifang knows their first-pass yield is 97.2%. They know that lot LOT-2026-1103 had elevated BLE signal variance and they flagged it during incoming inspection. They know the Thursday night shift runs 0.3% lower yield because a newer operator takes longer on the manual placement steps. They track all of this. It's their business.
But when the founder asks "how's production going?", they get an email: "Batch 044 complete, 480/500 shipped, on schedule." Three sentences. Not because the CM is hiding the other data. Because there's no shared system to deliver it. Collecting KPIs from contract manufacturers is difficult — teams must contact each CM and manually compile the data (Flexware). The CM's MES doesn't talk to the founder's spreadsheet. The test station logs sit on a local Windows machine. The rework records are in a notebook on the production manager's desk.
The CM is doing meticulous work. The founder just can't see it.
The switching trap
When quality problems appear and the founder can't see what's happening inside the factory, the instinct is to blame the factory. It's a reasonable instinct — if the product is failing and the only variable you can identify is the manufacturer, changing the manufacturer seems logical.
But the variable isn't the manufacturer. The variable is the information gap between the manufacturer and you.
Consider what actually happens during a CM switch. The founder transfers Gerber files, BOMs, and assembly instructions. Maybe a test specification, if one exists. The new CM does a DFM review, runs first articles, and starts production. Everything the new CM learns about yield, failure modes, and process optimization lives in the new CM's systems — just as invisible to the founder as the old CM's data was.
The founder traded one black box for another. The scenery changed, but the architecture didn't. The average cost of an electronics product recall exceeds $3.8 million (Ecosire) — switching CMs doesn't reduce that number.
Meanwhile, the knowledge the old CM accumulated over months of production — which component lots performed best, which reflow profile minimized tombstoning, which test points needed extra contact pressure — all of that walked out the door. The new CM starts from zero.
What the CM actually wants
Here's the part that surprises most founders: your CM wants you to see their data. Not because transparency is virtuous (though many CMs take pride in their work), but because visibility solves problems that cost the CM money.
Every time a founder emails "we're seeing field failures on batch 042," the CM's quality engineer has to stop what they're doing, dig through local files, correlate serial numbers with test records, compile a report, and send it back. This takes days. During those days, the founder is anxious and the CM is frustrated. Both are burning time on archaeology that could have been a dashboard query.
When there's a shared system, the conversation changes. The founder sees the yield data in real time. If something drops — say yield falls from 98% to 94% over three days — both sides see it simultaneously. The CM's quality engineer doesn't get an accusatory email a week later; they get a collaborative investigation on day one, when the root cause is still fresh and the affected units might still be on the line.
Shared visibility transforms the relationship from periodic audits and reactive blame into continuous collaboration. The CM gets credit for their 97.2% yield instead of being judged by the 2.8% that escapes. The founder gets confidence instead of anxiety. The email thread shrinks from daily status requests to occasional strategic discussions.
The infrastructure that's missing
The gap isn't talent. It isn't care. It isn't even process. The gap is a pipe — a shared system that makes the CM's existing data visible to the founder without requiring the CM to do extra work.
Think about what this pipe needs to carry: test results per serial number (not just pass/fail, but every measured value and the limits applied), stage progress (how many units have completed SMT, AOI, FCT, packing), yield metrics (first pass, retest rate, failure Pareto), and audit records (who ran which test, when, with what firmware version).
None of this data is new. The CM is already generating all of it. The problem is that it lives in local systems — MES databases, test station logs, operator tablets — with no standard way to share it outward. The founder's view of production is whatever the CM has time to compile into an email.
This is the infrastructure problem that AN & Factory solves. The test station integration captures results at the point of measurement and syncs them to a shared dashboard. The stage tracking updates automatically as units move through production. The yield analytics compute themselves from the test data that's already flowing. The CM doesn't change how they work — their test stations generate the same data they always have. The difference is that now a pipe exists to carry that data to the people who need it.
Same factory, different outcome
Imagine the forum post rewritten. Same founder, same CM, same BLE sensor. But this time, when the field failures appear, the founder doesn't write "is it time to switch?" Instead, they open the production dashboard, filter by the affected batch, and see that 94% of the failing units trace to component lot BLE-2026-1028. The test data shows those units passed FCT but with signal strength margins 2 dB lower than the lot average. The CM's incoming inspection had flagged the lot as within spec but near the edge.
The founder and the CM look at the same data. The decision is obvious: hold the remaining units from that lot, tighten the incoming inspection threshold for the next shipment, and work with the component supplier on consistency. No blame. No switching. No three-month qualification of a new factory. A problem identified in minutes, resolved in days, with both sides learning something that makes the next batch better.
The CM didn't change. The factory didn't change. What changed was the infrastructure between them — the shared pipe that turned invisible good work into visible, actionable data.
Your contract manufacturer is almost certainly better than you think. The question isn't whether they're doing good work. It's whether you've built the infrastructure to see it.
Ready to see what your CM already knows? AN & Factory builds the shared pipe between you and your manufacturer — real-time test data, yield dashboards, and stage tracking from the CM's existing workflow. Get in touch.
Sources
Have thoughts on this? Join the discussion on our forum.