The short answer: to judge whether a glass bottle decoration factory is stable, do not look at how long its equipment list is or how low its unit price is. Look at four things — whether first articles are approved, whether in-process inspection runs at a fixed frequency, whether process parameters are on record, and whether a problem can be traced to the specific person on that batch. A factory that does all four for real will almost certainly deliver reliably. A factory with the newest machines and none of the four will still come unstuck on a repeat order.
Four checks you can run on site
| Check | What good looks like | What to follow up with |
|---|---|---|
| First-article approval | A first-article verification record exists before start-up; no pass, no run | Where is the first-article approval sheet? Who signed it? |
| In-process inspection | Inspection at a fixed frequency, not "whenever there is time" | What is the frequency? Where are the inspection records? |
| Retained sample management | Samples are labelled and can be found by order | How long are samples kept? Which items are on the label? |
| Traceability | Batch, machine and responsible person can be identified | Can you look up an order from three months ago right now? |
One warning: this is not a question of whether the policy documents exist. It is a question of where the record is and whether it can be produced today. Putting a policy on the wall is easy; filling in every field of a sample label is hard.
Use numbers to find the real level
Qualitative descriptions are vague by nature, so ask in numbers.
The figures the company can state publicly are these: changeover time cut from about 1 hour to about 30 minutes; production problem frequency down from almost every day to 5-6 times a month; time to trace a defective product shortened from about 3 days to 5 minutes. Behind all three numbers is the same thing — the production process is recorded, so it can be reviewed and improved.
The clearest evidence is a real case: of a batch of 160,000 essential oil bottles sent to France, 2,000 were printed blurred. The system records identified the batch, the packing operator, the production time, the machine, the person responsible and the inspector within 5 minutes. That is what stability looks like when it can be verified — not a promise that nothing goes wrong, but the ability to locate the person within minutes when something does.
What to look at on the floor
- Defective product area: is there a separate isolated area, and is it clearly labelled? A factory that stores defective and good product together has, in effect, no floor management.
- Material labels: do semi-finished products carry the customer name, style, quantity and process status (to be printed, printed, to be packed, and so on)? Missing labels are a sign that material is about to be mixed up.
- Retained sample area: can a batch of samples be pulled out by order number on the spot?
- Data dashboard: does it show equipment running status, operator output, order progress and quality inspection records in real time?
- Customer-facing access: can customers check progress themselves? If they can track order progress and view production quality at any time through a mini-program, the data is genuinely connected rather than entered after the fact.
The company's structure is this: the new plant site at Shanggou covers 4,000 square metres and is divided into four workshops — automatic machine, manual machine, decal and packing — with 20 production lines across the plant, a team of about 30 people and daily capacity of 200,000 units. These figures can serve as a benchmark, but scale on its own does not equal stability. Scale has to be read together with the four checks above.
The cost of stability is hidden in the cost structure
A stable factory is not necessarily the cheapest, because record keeping, inspection rounds, retained samples and traceability all take labour. There is a useful way to tell: if a factory quotes far below its peers and cannot produce first-article records or retained samples, the cost it has cut is most likely the cost of quality management — and that cost will land back on you as rework and late delivery.
Conversely, if a factory can break its quotation down item by item (how the number of colours, the difficulty of the bottle shape, the quantity and the stacked processes each move the price), it understands its own cost structure, and that kind of factory is usually better at holding delivery dates.
A factory audit checklist for buyers
- Ask them to pull a first-article approval sheet on the spot.
- Ask them to pull the current day's inspection record on the spot.
- Name any order number from three months ago and see whether the retained samples can be found.
- Get the three numbers: changeover time, problem frequency and trace time.
- Check whether the defective product area is separate and fully labelled.
- Ask whether customers can check order progress themselves.
A factory that can deliver all six on the spot is stable. One that can only describe them verbally is hoping to be stable.