Direct answer. Reliability is not a verdict you reach about a supplier; it is a score you keep, updated once a year, on twelve measurable points. Six of them come from your own records — on-time rate, defect claims, response time, change handling, documentation and pricing behaviour. The other six come from public records and structured observation. Score each point 0 to 5, apply the weights that match your order profile, and a total below 30 out of 60 means start sourcing alternatives before the next buying cycle rather than after the next failure.
Why “reliable” has to become a score, not a feeling
Buyers decide whether to keep a supplier in one of two ways. The first is a feeling, usually formed during the crisis that nearly cost a season: the supplier handled it well, or did not. The second is a score, kept on paper, updated on a fixed date, and reviewed before the buying cycle rather than after it.
The feeling-based approach has a specific failure mode. A supplier that is excellent on nine points and catastrophic on the tenth looks fine right up until the tenth point is the one your programme depends on. This is how buyers end up in a relationship that reads as strong in every meeting and breaks at the worst possible moment — the score existed in nobody’s head, so nobody noticed the trend.
The scored approach has its own trap, which is the one this page is mostly about avoiding: scoring the things that are easy to observe and skipping the things that predict failure. Suppliers are energetic about the easy ones. Response time is easy to be good at. Capacity honesty during peak season is not, and it matters more.
The twelve points
Each point is scored 0–5, where 0 is a repeated failure, 3 is acceptable, and 5 is better than the industry norm. The evidence column matters as much as the point: a score that cannot be traced to a specific record is a feeling wearing a number.
| # | Point | What you score | Where the evidence comes from |
|---|---|---|---|
| 1 | On-time shipping rate | Percentage of orders shipped within the agreed window, across your last 12 months | Your own PO dates against bill of lading dates |
| 2 | Defect claim rate | Claims raised per 1,000 pieces, and the trend across four quarters | Your inspection reports and any customer returns attributable to the supplier |
| 3 | Response time on real questions | Time to a substantive answer — not an acknowledgement — on technical and commercial questions | Your own email log; measure from send to first usable reply |
| 4 | Change handling | How a mid-order change (substitution, quantity shift, size-mix revision) was priced and executed | The written change confirmation and the final invoice |
| 5 | Documentation discipline | Whether every shipment arrived with complete, correct and legible documents | Your customs broker’s exception log |
| 6 | Pricing behaviour | Whether quotes held for the stated validity, and how increases were explained when they came | Your quotation history; check whether the reason given was specific |
| 7 | Capacity honesty | Whether they told you a slot was full before you committed, or discovered it afterwards | Your booking log against actual production start dates |
| 8 | Process transparency | Whether inline and final inspection data was shared without being requested | Whether you received data or had to ask for it |
| 9 | Sub-tier visibility | Whether you know which fabric mill, which dyeworks, which trim supplier sits behind your order | Your own records — an honest gap here is a score of 1, not a mystery |
| 10 | Problem ownership | What happened the last time something went wrong: who proposed the fix, and who paid for it | The written record of your most recent claim |
| 11 | Continuity signals | Whether your merchandiser and QC contact have stayed in place, and how departures were handled | Your contact history; a revolving door is a leading indicator |
| 12 | Public record stability | Registered entity unchanged, no new enforcement records, annual filings current, no ownership change | Public corporate records for the contracting entity |
Twelve points, and only one of them — the last — requires no cooperation at all. That asymmetry is deliberate: eleven points measure the relationship, and one measures the entity that stands behind it. A supplier can be excellent at the first eleven and still be the wrong counterparty if the twelfth has quietly changed.
Scoring each point without guessing
1–2: the two hard numbers
On-time rate and defect claim rate are the only two points with an unambiguous answer, and they are the two buyers most often neglect, because calculating them means going back through a year of records. Do it once and the pattern is usually obvious. A supplier at 92% on time with a flat defect trend is different from one at 92% whose last two quarters were 78% and 85% — the average hides which direction the relationship is travelling.
Score the trend, not just the level. A supplier improving from 80% to 95% should score higher than a stable 96% only if the improvement is recent and unexplained; otherwise a stable number is the better signal, because it means the process is under control rather than being rescued.
3–4: response and change — where small suppliers separate
Response time should be measured on substantive answers. A 40-minute reply that says “let me check and come back” is worth less than a 24-hour reply that contains the answer, the reasoning and the alternative. Score the second reply, and note the first.
Change handling is the point where a supplier’s real cost structure becomes visible. Ask for a mid-order change and read three things: how long the answer took, whether the price impact came with a breakdown or as a round number, and whether the change was confirmed in writing before production moved. A supplier that absorbs a small change gracefully has margin; one that prices every change as a crisis is running tight, which is a fact about next year’s peak season.
5–6: documents and prices — the discipline points
Documentation failures are almost never fatal on their own and almost always a symptom. A supplier that sends a packing list without carton dimensions twice a year is careless. One that cannot produce a certificate of origin within the window your broker needs is a supplier whose export process is improvised, which means the same improvisation will show up somewhere more expensive.
On pricing, score whether quotes held for the validity they stated. A supplier that quotes 30-day validity and comes back in nine days with a higher number is telling you the first quote was not costed. That is not necessarily dishonesty; it often means the merchandiser quoted before checking with the mill. But it is a reliable predictor of how the next negotiation will go, and it deserves a low score rather than an excuse.
7–8: capacity and process — the two points that predict failure
If you score only two points on this list, score these.
Capacity honesty is measured by a single question, asked retrospectively: when your order was booked, was the slot real? Suppliers that overbook are not rare and are not always dishonest — the pattern usually reflects a sales team incentivised on confirmed orders and a production planner with no authority to say no. The consequence lands on you in the month before your season opens. A supplier that has told you “we cannot take that date” is demonstrating the exact behaviour you want, and it should score 5 for it, not 2.
Process transparency is measured by whether data arrives unbidden. Inline inspection results at 30, 60 and 90% completion, the needle detection record, the final inspection report — a supplier with a working process control system shares these because sharing costs nothing. A supplier that only ever reports good news, and whose reports appear only when asked, has a QC function that exists on paper. On our own orders we treat an unexplained absence of inline data as a defect of the supplier, not a gap in our records, because a factory running the checks it claims to run cannot avoid producing the output.
9–11: visibility, ownership, continuity
Sub-tier visibility is where trading partners and factories part company on capability. If your supplier cannot tell you which mill wove the fabric for your last order, then that supplier cannot investigate a chemical non-conformity, cannot run a second-source programme on your behalf, and cannot explain a shade variation that traces back to a yarn lot. This is a scoreable point, and a surprisingly large number of suppliers score 1 or 2 on it calmly, because nobody has ever asked.
Problem ownership should be scored on the most recent claim, not the best one. Read the record and answer four questions: who noticed the problem first, who proposed the remedy, whether the remedy was executed on the agreed date, and who paid. A supplier that pays for a rework it caused is worth more than one that never makes mistakes, because the second kind does not exist.
Continuity signals are the cheapest leading indicator on the list. If your merchandiser has changed three times in a year, the relationship you have is with a company, not a person, and every handover loses a little specification detail. Ask directly how the last handover was done — a proper one involves a written brief and a joint call. A two-line email saying “this is your new contact” tells you what the next quality dispute will feel like.
12: the public record, checked once a year
This is the only point you can score with no cooperation, and it is the one that most often changes between reviews. Four things to confirm for the entity named on your contract:
- The registered entity is the same one you contracted with — check the unified social credit code, not the trading name, which can be changed or used loosely.
- No new entry has appeared in the enforcement database administered by China’s Supreme People’s Court.
- Annual filings are current. A company that has stopped filing annual reports is a company in difficulty, and this is visible before it becomes visible in delivery.
- No change of legal representative, registered address or ownership that nobody mentioned to you.
The corporate registration records are held by China’s State Administration for Market Regulation at samr.gov.cn, and enforcement records are searchable at zxgk.court.gov.cn. Ten minutes once a year.
Weighting the score for your order profile
The twelve points do not carry equal weight, and the right weights depend on what your programme actually needs. Three profiles cover most buyers; pick the one that matches or adjust the columns.
| Point | Growing brand, small volumes | Established retailer, seasonal peaks | Regulated market, compliance-heavy |
|---|---|---|---|
| 1 · On-time rate | ×1 | ×3 | ×2 |
| 2 · Defect claim rate | ×3 | ×3 | ×2 |
| 3 · Response time | ×3 | ×1 | ×1 |
| 4 · Change handling | ×3 | ×2 | ×1 |
| 5 · Documentation | ×1 | ×2 | ×3 |
| 6 · Pricing behaviour | ×2 | ×2 | ×1 |
| 7 · Capacity honesty | ×2 | ×3 | ×2 |
| 8 · Process transparency | ×2 | ×3 | ×3 |
| 9 · Sub-tier visibility | ×1 | ×2 | ×3 |
| 10 · Problem ownership | ×3 | ×2 | ×2 |
| 11 · Continuity | ×2 | ×2 | ×1 |
| 12 · Public record | ×2 | ×2 | ×3 |
A small brand with no buying power is not optimising for price; it is optimising for a supplier that answers, tolerates changes and takes a small order seriously. A retailer fighting for a fixed launch window is optimising for the date. A compliance-heavy market is optimising for the paper trail, because the cost of a documentation failure there is measured in customs, not in customer complaints.
What the total tells you
Weighted maximum is 60. The thresholds below are starting points; the useful discipline is applying the same thresholds every year rather than renegotiating them in the meeting.
- 48 and above — invest in the relationship. Consider moving volume their way, and consider whether they should be the supplier you take into a new market first.
- 36 to 47 — stable, keep going, and work the two lowest points with a specific request rather than a general complaint.
- 30 to 35 — watch closely and start the alternative-supplier search on a low priority. Do not place a first order for a new market or a new category with this supplier.
- Below 30 — start sourcing alternatives before the next buying cycle. A total below 30 with a falling trend over two reviews is a relationship to end politely and early, while you still have time to transition.
Two scoring rules matter as much as the thresholds. First, a score of 0 on point 10 (problem ownership) or point 12 (public record) caps the total, regardless of the other points — a supplier that will not own a problem and an entity whose registration has changed are both disqualifying in their own way. Second, track the direction. A 40 this year against a 46 last year is a worse position than a 36 that came from a 28, and the numeric total alone will not tell you which one you are looking at.
Running the review without telling the supplier
Nine of the twelve points come from records you already hold, and the other three come from public registers. You can complete a full review without the supplier knowing it is happening, which is worth doing at least once, because a supplier that knows it is being scored will behave differently for a quarter.
When you do share the result, share the two lowest points and a request, not the total. “Your documentation has cost us two customs delays this year; can you confirm who prepares the paperwork and when” is actionable. A score out of 60 is not — it invites an argument about the scoring rather than a change in behaviour.
When to replace rather than repair
Replacement is expensive: a new supplier means new lab dips, new size-set approval, new paperwork, and a first order that carries more risk than any repeat will. That cost is why the repair path is usually right. Three situations are exceptions:
- A disqualifying score. Zero on problem ownership, or a changed legal entity behind your contract, or a migration of production to a plant nobody will name.
- Two consecutive reviews below 30. The trend has been measured twice, which means it is a pattern rather than a bad quarter.
- A structural mismatch that no score will fix. If your programme needs a compliance file naming the manufacturing site and your supplier cannot provide one, the relationship is not underperforming — it is the wrong shape.
When you do replace, run the transition deliberately: overlap the two suppliers for one order, keep the outgoing supplier responsible for the last delivery they committed to, and keep the incoming one on a smaller first order than your programme needs. Our buyer’s guide to choosing a manufacturer covers the selection side, and the verification checks that distinguish a factory from an intermediary apply to the incoming candidate before you commit volume.
Compliance and verification
The points above lean on records and registers rather than on impressions, and the registers are public:
- Entity identity and stability — corporate registration, annual filings and scope of business, administered by China’s State Administration for Market Regulation (samr.gov.cn).
- Enforcement history — the Supreme People’s Court enforcement information database (zxgk.court.gov.cn).
- Social compliance — amfori BSCI audits, held by the manufacturing site that is audited, not by a sourcing company acting for it.
- Chemical compliance — OEKO-TEX Standard 100 certificates name an article and a licence holder; the holder can be checked by certificate number through the issuing body’s label check.
- Quality evidence — pre-shipment inspection to a documented sampling plan under ISO 2859-1, with the report issued by a named independent body rather than by the production site, the distinction covered in our guide to reading an inspection report.
- Trade context — country-level sourcing context from the US Commercial Service China Country Commercial Guide, shipment-level statistical data from UN Comtrade, and the GS1 standards that make carton-level traceability possible.
Next step
If you want the scorecard as a working file, tell us your order profile — market, styles, annual volume and whether you are buying on a launch date — and we will send the weighted version with the thresholds set for that profile. If you would rather run it on us first, ask for the evidence behind points 7, 8 and 9 on your current order, and score the answers yourself.
· Xiamen Unitex Trade Co., Ltd. — OEM & ODM underwear sourcing partner