Short answer: do not start a silicone supplier evaluation with "how big are they, what is their reputation, who have they worked with" — start with what can be measured and verified. Each of the seven criteria below is checkable on the spot: either there is a document, or there is a specific number, or it can be turned into a contract clause. Score them on a simple sheet; that beats two hours of impressionistic conversation.
The value of these criteria is not "picking the best factory" but surfacing risk early. Most sourcing disputes — slipping lead times, production parts that feel different from the samples, test reports that cannot be traced — show warning signs before the order is placed; they simply were never quantified. One caveat: the benchmarks below are negotiation reference points, not mandated industry standards — adjust them to your product’s risk level.
| Criterion | How to quantify it | Suggested benchmark | How to verify |
|---|---|---|---|
| Dimensional pass rate | First-article full dimensional report plus batch sampling defect rate (or CPK on critical dimensions) | Batch defect rate ≤2%; CPK ≥1.33 on critical dimensions | Request the original first-article report; write the AQL sampling level into the contract |
| On-time delivery rate | Share of orders delivered on time over the last 6–12 months | ≥95% | Ask for disclosable delivery history; if unavailable, agree late-delivery penalties in the contract |
| Sampling-to-production consistency | Whether samples and first production parts share the same process, line and tooling | All three must match | Ask directly which tool is used for sampling and which for production; vague answers are a red flag |
| Verifiability of test reports | CNAS/CMA accreditation seals, whether the report number is traceable on the issuing lab’s website, and coverage of target-market regulations | Valid seals, traceable number, test items cover the target market | Verify the report number on the issuer’s site or by QR code; check the tested sample matches the model you are buying |
| Material traceability | Whether a batch number traces back to compound grade, supplier and production date | Complete batch records, traceable backwards | Pick a random completed batch number and ask them to trace it on the spot |
| Tooling ownership and life | Who owns the tool, warranted shot count, who pays for re-tooling at end of life | Stated explicitly in the contract; verbal assurances are not acceptable | Review the contract clause by clause; confirm transferability and scrapping conditions |
| Defect response and remediation | Defect criteria, replacement deadline, who bears cost and freight | All three quantified in writing | Check for concrete days and amounts; wording like "to be resolved through discussion" is effectively no clause at all |
Four Typical Scenarios
Is comparing quotations enough?
No. A quotation reflects only price, and only under the assumption that everything goes smoothly. What actually separates suppliers is who absorbs the loss when things go wrong: late delivery, production that differs from the sample, failed testing. When comparing quotes, build a table that includes the seven criteria alongside price — at the same unit price, a supplier with 98% on-time delivery and written penalties is a very different long-term cost from one who cannot produce the data.
How do you audit without visiting?
Run a small trial order end to end — the cheapest and most honest verification there is. Place a 50–500 piece order and take it through enquiry, sampling, approval, production and after-sales, watching three things: was sampling on time, do production parts match the samples, and how fast do they respond when something goes wrong. A site visit shows static capability; a trial order shows dynamic behaviour, and the latter predicts the relationship far better. A video walkthrough helps confirm the equipment and workshop actually exist.
How can you tell whether a test report is genuine?
Three steps. First, check the accreditation seal: domestic reports should carry a CNAS or CMA seal; without one the report carries little weight. Second, verify the report number on the issuing lab’s website or by scanning the QR code — it only counts if it resolves, and many buyers skip this. Third, confirm the tested sample is the model you are actually buying — the most common trap: the report is genuine but covers a different product, colour or batch. Also check validity: once the compound or formulation changes, the old report no longer applies.
Does a bigger factory mean a more reliable one?
Not necessarily — for small-volume buyers it is often the reverse. Large factories bring capacity and systems, but scheduling priority usually follows order size, so your 500-piece order may sit behind a major client. Mid-size and smaller plants are frequently better on sampling speed, design flexibility and willingness to take small orders. The real question is not size but where your volume sits in their customer mix — a buyer of a few thousand pieces a year tends to be deprioritised at a plant built for million-piece runs.
Supplier Evaluation Checklist
- Set the product risk level first: food contact, medical and baby products warrant the strictest checks
- Request first-article dimensional reports and agree an AQL sampling level for production
- Ask for disclosable on-time delivery data; if unavailable, agree late-delivery penalties
- Ask outright whether sampling and production share the same process, line and tooling
- Verify test reports by number; confirm the tested model and validity period
- Pick a random batch number to test material traceability
- Confirm tooling ownership, warranted shot count and re-tooling responsibility clause by clause
- Write defect criteria, replacement deadlines and cost allocation as quantified clauses
- Run a small trial order end to end before scaling up
FAQ
Should I go with the cheapest quote?
You can, but find out where the low price comes from. Common sources: a different process (compression instead of injection, or skipping post-curing), a lower compound grade (regrind or a lesser specification), or omitted testing. All three convert into risk somewhere you cannot see. The test: ask them to break the quote into compound grade, process route and test items. A low price that survives that breakdown is genuinely cheap; one that does not has usually left something out.
Samples were great but production feels different — is that a defect?
It depends on whether you are within agreed tolerances, but most cases reflect a process change rather than simple variation. The usual cause is sampling on a soft tool with manual trimming while production runs on automated steel tooling — flash, density and surface feel differ naturally. Compound batch or cure-condition changes are second. There is only one way to prevent it: state in writing before ordering that sampling and production use the same process, line and tooling, and require the first production article and the approved sample to be sent together for record. Once the goods arrive, only the contract’s acceptance clause can help.
Should suppliers be re-evaluated every year?
At least annually is sensible, but do not confine it to year-end. A more practical approach is to fold the seven criteria into daily work: record delivery punctuality and sampling defect rate per batch, and roll them up quarterly. Year-end review then rests on real data rather than recollection. Watch for two triggers that warrant immediate re-evaluation — the supplier changes compound source, or a test report expires without renewal. Even long-standing partners should be re-checked when either occurs.
LXYSILICONE can supply verifiable documentation against each criterion above: first-article dimensional reports, AQL batch sampling records, third-party test reports verifiable by report number, and batch-traceable compound records. Standard ranges start at 50 pieces, and trial orders can be scaled up progressively.