Looking for an OEM/ODM Case Partner?
The supplier that answers your RFQ fastest is rarely the one worth signing. Between first enquiry and mass production there are six stages, and each one has a defined deliverable and an acceptance test. Where the test is vague, every later stage degrades into a verbal promise. This walks the chain from a buyer's side: what belongs in the RFQ pack, what sampling really proves, why tooling ownership must be settled before T1, how first article and ramp-up are gated, and how two rounds of screening remove the wrong partner early instead of three months into production.
01Stages 1–2: A Thin RFQ Guarantees Rework Later
Stage one is clarification. The RFQ pack needs five things: service conditions and load, internal dimensions within ±0.3 mm, annual and first-order volume, cosmetic grade including where sink marks and flow lines are tolerated, and packing plus shipping method. Leave one out and the supplier fills the gap from habit. Whatever they guess wrong returns to you later as a change cost or a slipped date.
Stage two is DFM and quote structure. Ask for seven lines, not one total: tooling, material per part, machining, hardware, foam, packing, tooling amortisation. Material is typically 35% to 45% of total cost, so this line has to be defined the same way on both sides. Add a manufacturability note naming which wall sections will sink, which radii cannot be moulded, whether draft is sufficient. A supplier willing to write those three points usually warns you early at every later stage too.
02Stage 3: Sampling Tests the Process Window, Not the Part
Sampling takes 15 to 25 days. The deliverable is not one part but three to five samples, a critical dimension report, resin grade and lot number, and the moulding parameter sheet, with the dimension report due inside 48 h. Acceptance means dimensions inside drawing tolerance, cosmetics at the agreed grade, assembly without interference, and a parameter sheet that can be repeated. One good part proves nothing.
The question worth asking here is what they would change if this batch were wrong. An answer of we will adjust the settings suggests no process window exists in their heads. Someone who names mould temperature, hold pressure or gate location is doing engineering. Tool changes cost least at this stage, so this is where to spend your arguments.
03Stage 4: Settle Tooling Ownership Before T1, Then Only Discuss Limits
Tooling runs 35 to 55 days and T1 usually takes 3 rounds: fill and cosmetics first, dimensions and warpage second, cycle and stability third. Deliverables are a trial report per round, parts, and a defect list with closure status. The gate is 30 consecutive shots with no critical defect in round three, plus enough data to calculate capability on key dimensions at a settled 60 s to 90 s cycle.
Ownership belongs in the contract before T1: whether tooling cost sits inside the piece price, who holds title, how the tool is disposed of after the programme ends, who authorises modification. Write the modification limits too, splitting what the supplier may change alone (gate, ejectors, cooling lines) from what needs written approval (cavity, sliders, parting line). Unclear limits turn every change into an argument.
04Stages 5–6: First Article in 24 h, Then Three Ramp Batches
The first article report is due within 24 h and covers four things: full dimensions, appearance, assembly, material certificate, with moisture below 0.2%. Acceptance means all dimensions pass, critical ones keep 0.05 mm to 0.1 mm of tool correction allowance, and assembly shows no interference. No signature, no production start. Once that rule bends, every later inspection is only damage control.
Ramp-up runs in 3 batches: a small lot for stability, a second for cycle time and defect distribution, a third at full rate. Each shipment is sampled at AQL 1.0 on critical items and AQL 2.5 on general ones, and ramp-up closes only when the defect rate falls across all three. At 500 to 1000 units MOQ, that sequence covers the tool's wearing-in period almost exactly.
05Two Screening Rounds: Paper First, Downtime Second
Round one needs no plant visit. Request four documents: the DFM note, the cost breakdown, a trial report template, and a stage plan from a comparable project. If two of the four are boilerplate with no figures in them, stop there. This round removes roughly half the candidates for the price of a few emails and two calls.
Round two is half a day on site, and the point is not equipment. It is how they handle abnormality. Ask three questions: how was the last bad batch caught, what caused the longest stoppage, and and at how many shots tool maintenance is scheduled, whether 5,000 or 10,000. Answers with times, stations and names indicate records exist. A flat no problems either means none occurred or none will be shared.
| Sampling lead time | 15–25 days |
|---|---|
| Tooling lead time | 35–55 days |
| T1 trial rounds | 3 rounds |
| First article report | within 24 h |
| Ramp-up batches | 3 lots |
| Minimum order | 500–1000 units |
| Tool design life | 300k–500k shots |
| Acceptance sampling | AQL 1.0 / 2.5 |
FAQ
Is a 30% gap between two quotes normal?
Not against the same RFQ. A gap that wide usually means different bases: theoretical weight versus weight including runners, tooling amortised versus billed separately. Flatten both into the same seven lines before comparing totals, or you are pricing two different products.
Must the buyer own the tooling?
Not necessarily, but title and disposal rights must be written down. If the supplier holds it, agree the handover process and cost at end of programme. If you hold it, agree storage location and maintenance duty. Both models work; ambiguity does not, because disputes surface on the day the line stops.
Why does a good sample still fail in production?
Sampling often runs a single cavity or a light load. Production is about continuity and cooling consistency. Watch three things: mould temperature stability, whether cycle time gets squeezed, and whether material changes come with a proper purge. All three tend to be relaxed during sampling.
How do you spot a supplier hiding problems?
Listen to the resolution of the answer. Naming a station, a time and a responsible person suggests a system. Repeated reassurance with no detail suggests either no system or no intention of sharing it. Then ask for a three-month defect Pareto and check whether the story matches.