From Patent Filing to CE Mark in 11 Months: A Turnkey OEM Email Thread

There is a folder in this inbox called "Programs the next founder should see." This thread is at the top of it. The founder — referred to here as Marta, at her own request — runs a six-person hardware startup in Utrecht. In January 2025 she signed an NDA with us, wired a tooling deposit, and sent the email that starts this thread: "We need to ship for Christmas. Tell me honestly if that is possible." Eleven months later, her first container cleared Rotterdam on November 16, twelve days before the retail blackout date her distributor had imposed.

What follows is the thread, abridged and anonymized but otherwise real, with commentary between the emails. The boring messages have been kept on purpose, because the boring messages are where programs actually live. If you are a first-time hardware founder, read this the way you would read a flight recorder: not for the drama, but for the exact moment where the schedule almost died, and what brought it back.

Week 0: The question every founder asks, and the answer nobody likes

From: marta@founder.eu · To: project@jinlong · Tue Jan 7, 09:14 CET

We have a working prototype, a provisional patent filing, and a retail commitment contingent on November delivery. Our industrial designer says design freeze is realistic by end of February. We have been quoted everything from "six months, easy" to "eighteen months if you are lucky" by other suppliers. We need an honest number. Can we ship a CE-certified product in November, and what has to be true for that to happen?

Marta's question is the one every founder asks, and the honest answer is the one nobody likes: a November ship date from a January start is achievable, but only if six specific things go right, and only if the two or three things that go wrong are recovered inside the month they happen. That was the message on the first call, and then it was put in writing, because a verbal critical path is worth nothing at week 30.

From: project@jinlong · To: marta@founder.eu · Wed Jan 8, 16:40 CST

Honest answer: yes, with conditions. Here is the critical path as a chain, not a task list. Design freeze Feb 28. Mold first steel Mar 14. T1 tryout Apr 25. T2 May 16. EMC pre-screen Jun 20. Technical file to lab Jul 4. CE certificate Sep 5. Mass production Oct 17. First container Nov 4. Every date here is the latest acceptable, not the target. The two places programs die are EMC and the mold tryout loop, so pre-screening and a second-cavity contingency are in the plan from day one. If any milestone slips more than two weeks, you will hear the same week, with the recovery plan attached. That is the deal.

Notice what the PM's answer does. It gives dates as latest-acceptable rather than targets, it names the two failure points in advance, and it commits to same-week disclosure of slips. Because a hardware founder's scarcest resource is not money but calendar, the most valuable thing an OEM can offer is not optimism but early bad news. This is the first thing every founder is told to look for in a project manager, and it cannot be verified from a website. It can only be verified by watching how the PM behaves the first time something breaks.

Weeks 6-16: The mold that failed its first tryout

Design freeze happened on February 26, two days early, which almost never happens and which is mentioned here only because it created the buffer that saved the program later. Molds cut first steel on March 13. Then, on April 24, the email every program dreads arrived, one day ahead of the scheduled T1 tryout.

From: project@jinlong · To: marta@founder.eu · Thu Apr 24, 11:02 CST

T1 result is in, and the bad news comes first. The main housing passes dimensionally on 41 of 44 critical dimensions. The three failures are all on the dust-cup latch boss: it is 0.18 mm undersize at the snap feature, and the snap will not survive the 10,000-cycle endurance test at this dimension. Root cause is shrinkage anisotropy in the glass-filled nylon, which was under-predicted on this gate location. Recovery plan: the gate is adjusted and 0.22 mm of steel is added at the boss, which costs 8 working days, not 10 working days of recut. T2 moves from May 16 to May 27. The rest of the critical path is unchanged because the buffer was built at design freeze. This is not the news anyone wants at T1; the trade-off is that finding it here is far cheaper than finding it at endurance test.

This email is why the program survived. Because the mold shop and the project manager sit in the same building, the failure was measured, root-caused, and scheduled for repair within twenty-four hours of the tryout. The same failure has been watched take three weeks on programs where the mold shop is a subcontractor two provinces away, because the tryout report travels by email, the corrective plan waits for a commercial negotiation about who pays, and the recut waits for the subcontractor's machine queue. The eight-working-day repair in this thread is not a credit to anyone's heroism; it is a structural property of an in-house tool room with 1,000 mold sets per year of throughput and 5-axis machines that were idle on a Thursday morning.

There is also a harder truth in this email that founders should notice. The latch boss failure was not caused by a bad mold maker; it was caused by a material behavior that is genuinely difficult to simulate perfectly on a first article. Glass-filled nylon shrinks differently along and across the flow direction, and a snap feature is the least forgiving geometry for that anisotropy. The plan did not assume the mold would be perfect. The plan assumed something would fail, and priced the recovery into the schedule. That is the difference between a timeline that survives contact with reality and one that does not.

Week 24: The EMC failure, and the week that decided the program

If the mold was the predictable scare, EMC was the existential one. On June 18, two days before the scheduled pre-screen, the PM sent a one-line holding email: "Pre-screen moved to Friday. I will call you after, whatever the result." Then, on June 20 at 19:47 China time, the call came, and at 21:15 the follow-up email.

From: project@jinlong · To: marta@founder.eu · Fri Jun 20, 21:15 CST

Pre-screen result: fail, by 4.2 dB on the radiated-emissions peak at 156 MHz, which is the motor drive's switching harmonic coupling onto the battery lead loom. This is the failure warned about in January, so do not panic; this exact signature has been seen three times. The fix is a split-ferrite on the battery lead plus a ground-strap reroute, both of which are mechanical changes, not board respins. The modified harness can be in the lab Tuesday and the re-screen Thursday. If Thursday passes, zero days are lost on the critical path because the technical file does not need the EMC report until July 4. If Thursday fails, three weeks are lost and that will be on the call Thursday night. Go to sleep on this side; someone will be awake on the other side.

This email has been read perhaps twenty times, and the line that still matters most is the one about sleeping on each side of the time difference. It is not bravado. It is the specific commitment a founder is buying when she chooses a turnkey OEM over a cheaper assembly shop: a project manager whose calendar is fused to hers across an eight-hour time difference. The re-screen passed on Thursday, June 26, by a margin of 2.8 dB, and the critical path lost zero days.

Here is the structural point, because founders always ask whether the EMC failure could have been prevented entirely. Probably not, and here is why. A brushless motor switching at high frequency is an RF transmitter whether it wants to be or not, and the way its harmonics couple onto cable looms inside a housing depends on geometry that simulation approximates but does not perfectly predict. Because the coupling path is geometry-dependent and the geometry is only final once the real harness meets the real housing, the EMC pre-screen is the first moment the full system exists as an electrical object rather than a model. Programs that die at EMC are not the ones that fail the pre-screen; they are the ones that skip the pre-screen and fail at the notified body's accredited lab, where each revision loop costs three weeks and a re-test fee. An in-house pre-screening capability, which is part of the QA organization at the Yuyao campus, is what turns a potential six-week catastrophe into a six-day scare.

Weeks 32-46: The quiet months where programs are actually won

The rest of the thread is, by design, boring, and a sample is included precisely because it is boring. The technical file went to the lab on July 3, a day early. The CE certificate arrived on September 2, three days early. Mass production started October 15. The first container shipped November 4.

From: marta@founder.eu · To: project@jinlong · Tue Nov 4, 22:31 CET

Container is on the water. My distributor has confirmed the November 16 receiving slot. I just reread your January 8 email with the critical path, and the thing that strikes me is that five of eight milestones were hit within three days, and the three that slipped were the three named in advance. Eleven months have been spent waiting for the disaster that never came, and only now is it clear that the absence of disaster was the product. Thank you. The second SKU starts in January, and the first email has already been drafted.

Marta's last line has lived in the email signature for internal training ever since, because it names the thing a turnkey OEM actually sells. The absence of disaster is the product. The mold that failed and was repaired in eight days, the EMC spike that was found and fixed in six, the forty-six weeks of Tuesday-morning replies — none of these appear on an invoice line, and all of them are the reason a six-person startup in Utrecht has a product on a retail shelf this Christmas instead of a lesson about offshore manufacturing.

What would have made this program fail?

Three things would have killed this program, and all three are choices a founder makes at supplier selection, not during execution: a subcontracted mold shop, a supplier without in-house EMC pre-screening, and a project manager whose incentive is to hide slips rather than disclose them. Each of these deserves a moment, because they are the real content of this article and everything above is the evidence.

A subcontracted mold shop turns every tryout failure into a commercial negotiation before it becomes an engineering task. The eight-day repair in week 16 becomes a three-week one, and two such failures end the Christmas window. A supplier without in-house EMC pre-screening discovers the 156 MHz spike at the accredited lab, where the re-test queue alone is three weeks, and the program dies in September with a certificate it cannot get. And a project manager who hides slips converts every recoverable problem into a fatal one, because recovery time is the one thing that cannot be bought back once it is spent.

Because a founder cannot audit any of these three things from a quotation, the only reliable signal is behavioral: how the supplier talks about failure in the first meeting. A supplier who tells you in January which two milestones will probably slip is a supplier who will tell you in June when they do. A supplier who promises that nothing will slip is telling you that you will learn about the slips from your distributor. This is the advice now given to every founder who asks, and it is worth more than any factory tour.

Which OEM model should a startup choose?

Choose a full turnkey OEM with in-house mold, motor, and assembly when there is a single platform, a hard retail window, and no internal engineering team of your own. Choose an assembly-plus-test OEM when the motor and mold supply are already owned and only integration is needed. Choose a pure contract assembly shop only when the product is mature, the engineering staff is in place, and tolerance for managing the critical path in-house exists. Marta's program needed the first model, and the second SKU, which starts in January, will use the same one, because the thing she bought was not manufacturing capacity but calendar certainty.

For any founder holding a prototype and a November deadline right now, the exercise worth doing is simple. Write the critical path as a chain of dependencies with latest-acceptable dates, name the two milestones most likely to slip, and ask the prospective OEM to do the same in writing. The supplier who answers that email concretely within one working day is the supplier who will answer the June 20 EMC email at 21:15 on a Friday. And that, more than any price in any quotation, is what decides whether the product is on a shelf at Christmas.

Frequently asked questions

References and site links

External references back the certification and timeline figures cited above:

• Notified-body identification and scope for EU conformity assessment: the European Commission's NANDO database, which is where every lab that touches a technical file has its accreditation verified.
• Lithium cell transport testing that applies to the cordless variant: the IATA Dangerous Goods Regulations, which cover lithium battery transport by air, sea, and land under the UN Model Regulations framework.
• Motor efficiency and test conditions referenced in the EMC discussion: IEC 60034 via ISO.
• Quality-management traceability that underpins the same-day failure-disclosure practice: ISO 9001 quality management systems.

Site pages referenced in this thread:

• Vacuum cleaner OEM product page, including the flexible stick platform this program shipped.
• About Jinlong, with the tool room, QA organization, and assembly floor that the critical path ran through.
• Contact, for the founder holding the next January email.

Why publish a client's email thread, and about the author

Because founders ask for references and then ask the references what actually went wrong, the failures and recoveries are shown directly rather than hidden behind a curated list of programs where nothing slipped. Because Marta gave permission after reading the final draft, the thread is real rather than reconstructed from memory, which is why the dates and the dB margins are exact. Because the next founder deserves to see the mold failure and the EMC scare before signing the tooling deposit, the difficult emails are included on purpose. If the next program looks calmer than this one, one of two things is true: either the product is genuinely simpler, or the bad news has not arrived yet.

About the author

Wanchen Xuan is a foreign trade specialist at Ningbo Jinlong Electric Appliance Co., Ltd., and is usually the person on the project@jinlong side of threads like this one. She runs turnkey vacuum cleaner programs for European and North American brands from the Yuyao campus, and continues to believe that same-week bad news is the most valuable thing an OEM sells. Connect on LinkedIn.