![[AN & Factory]](/static/img/logo.svg){kind=logo}
The Real Cost of Skipping Factory Testing
Three weeks after your Kickstarter units ship, the emails start. "My device won't pair." "Screen flickers after ten minutes." "Worked for a day, now nothing." You shipped 3,000 units. You don't know if you're looking at five bad ones or five hundred. And every morning brings another inbox full of one-star reviews from backers who believed in you enough to pay up front.
This is what skipping factory testing actually costs. Not the dollar amount you "saved" on the line — the dollars, hours, and trust you'll burn trying to recover.
The math nobody does until it's too late
Functional circuit testing costs between $0.50 and $2.00 per unit to run (Promwad). For 3,000 units, that's $1,500 to $6,000. The fixture itself runs $3,000 to $8,000, amortized across every batch you'll ever build on that design. Call it $5,000 all-in for your first production run.
Now here's what happens without it. Industry data consistently shows a 2–4% defect escape rate for boards that pass visual inspection and in-circuit testing but skip functional testing (Predictable Designs). These are the defects that look electrically correct but don't work as products — a sensor that reads 15% low, a radio that initializes on the bench but drops connection under thermal load, a voltage regulator that oscillates at frequencies your multimeter can't catch.
Take the middle of that range: 3% of 3,000 units. That's 90 devices that will fail in your customers' hands.
Each return costs you real money. Two-way shipping: $12–$25. Customer support time to diagnose and process: $25–$50 per ticket. Replacement unit or refund: the full product cost. Diagnosis time at your end: 15–30 minutes of engineering time per unit. Add it up and a single field return costs $80 to $200 depending on your product price and where your customer lives.
At 90 returns and $80 each, you're at $7,200 in direct costs. At $120 each — more realistic for international shipping — that's $10,800. You spent nothing on testing to save $5,000, and the field failures cost you twice that before you count the damage you can't invoice.
The costs that don't show up on a spreadsheet
The direct return costs are the easy part. The hard part is everything else.
Negative reviews compound. A Kickstarter project that ships 3,000 units with 90 failures will collect 30–50 public complaints within the first month. On Amazon, each one-star review significantly impacts the algorithm, and products that exceed category-average return rates can face listing suppression or the "Frequently Returned Item" badge (Riverbend Consulting). A product that launches at 4.6 stars and drops to 3.8 in the first month loses organic visibility — and with it, the sales velocity that makes hardware economics work.
Amazon's return rate thresholds are particularly punishing. Products that exceed category-average return rates get flagged, and consistently high returns can lead to listing suppression. For electronics, the average return rate sits around 8–12%. If your defective units push you above that, your listing starts to disappear from search results.
Then there's the support burden. Those 90 failing devices don't announce themselves all at once. They trickle in over weeks and months, each one consuming a support ticket, an email thread, a troubleshooting session. Your two-person team is spending 20 hours a week on returns instead of working on the next product revision. For two months, your roadmap stalls because you're running a reverse logistics operation you never planned for.
And the warranty reserve. If you're raising a Series A, your investor will ask about warranty liability. "We don't have test data" is not the answer that closes a round.
The Kickstarter reckoning
A real scenario, composited from several hardware startups we've talked to. A team designs a BLE wearable — an accelerometer, a microphone, a Bluetooth radio, a battery management circuit. Their CM in Shenzhen assembles the boards, runs AOI, does a basic power-on check: LED lights up, device shows up on Bluetooth scan. "100% tested," the CM reports. And they're not lying — every unit powered on and blinked.
But a power-on check isn't FCT. Nobody measured the microphone sensitivity. Nobody verified the accelerometer reads within spec on all three axes. Nobody checked that deep-sleep current draw stays below 15 microamps. Nobody ran a 30-second BLE connection stability test.
Of the 3,000 units that shipped, 4% had a microphone issue — a component lot where the MEMS element had slightly lower sensitivity, enough to pass a "does audio come through" check but not enough to work reliably at arm's length. Another 1% had accelerometer drift from reflow thermal stress. Total: about 150 units that worked on the factory floor but failed in the real world.
The team spent the next eight weeks doing nothing but fielding returns, debugging, negotiating with the CM, and shipping replacements. The CM, to their credit, offered to retest remaining stock — but by then the inventory was already on customers' wrists. There was no way to identify affected units without serial-number-level test data, because there was no serial-number-level test data.
Total cost: roughly $18,000 in direct returns and replacements, plus two months of engineering time diverted from the V2 product, plus a Kickstarter comment section that reads like a product obituary. The FCT setup that would have caught everything? About $8,000.
Why your CM wants you to test
Here's the part that surprises most founders: your contract manufacturer wants you to invest in FCT. Good CMs hate field returns as much as you do — maybe more, because every complaint lands on their quality team's desk too.
When a founder defines clear test specs and provides functional test fixtures, the CM's job gets easier, not harder. They have unambiguous pass/fail criteria instead of judgment calls. Their operators run a defined sequence instead of improvising. Their quality team can point to test data when questions arise. And their rework rate drops, because FCT catches assembly issues early enough to fix them on the line instead of after boxing and shipping.
The CMs who push back on functional testing aren't hiding anything — they're responding to incentives. Testing costs time, time costs money, and without a clear spec from the founder, the CM is guessing at what "tested" means. When you hand them a test specification, a fixture, and clear pass/fail limits, most CMs will run it gladly. You've removed the ambiguity that makes testing feel like a liability instead of a quality gate.
The $5,000 insurance policy
Factory testing is the cheapest insurance a hardware startup can buy. A basic FCT station — a pogo-pin fixture, a programmable power supply, a test sequencer — costs $5,000 to $15,000 (FixturFab), less than one month of a junior engineer's salary. It pays for itself on the first production run. Every run after that, the per-unit cost drops while the protection stays the same.
But the real value isn't the returns you prevent. It's the data you create. Every unit that passes FCT generates a record: serial number, timestamp, every measured value, every limit, pass or fail. That record is the difference between "we think our products work" and "here's the proof." It's the difference between a full recall and a targeted one. It's the difference between a customer support nightmare and a five-minute serial number lookup.
The founders who skip FCT aren't saving money. They're borrowing it — at 25 to 100 times the interest rate — from their future selves, their customers, and their brand. The invoice always comes due. It just arrives disguised as support tickets, return labels, and one-star reviews.
Every unit you ship without functional test data is a unit you can't defend, can't trace, and can't learn from. The factory floor is the last place you control the outcome. After that, it's your customer's problem — and then it's yours.
Ready to set up factory testing for your product? AN & Factory provides test station software, signed test records, and CM integration — from test specification to fixture design to production monitoring. Get in touch.
Sources
Have thoughts on this? Join the discussion on our forum.