One bottle, three samples一个瓶子，我们打了三次样

◆ 序幕

Some orders you remember for a long time. This one is memorable because of a single bottle — and three attempts at it.

The client is a European skincare brand. On the first video call, their procurement director said something very simple: "This bottle must not leak when it ships on its side."

That was all. No specification sheet, no test standard, no reference sample. The bottle was 30ml square-shouldered glass with a screw-on dropper cap — a format we had made many times.

So my judgement at the time was: this is not a difficult job. That judgement turned out to be too optimistic.

◆ 第一幕

For the first sample we used our most proven approach: a standard threaded neck with a flat gasket. It is the most common sealing solution in the industry — low cost, simple to assemble, adequate for most products.

We sent it. The client's reply was two words: not acceptable.

The problem was not that the work was rough. There is nothing wrong with a flat gasket. The problem was that we had assumed the client's use case — we were thinking "no seepage once the cap is tightened", and the client was thinking "the bottle lies on its side in a courier box for three days, gets knocked about the whole way, and still must not leak".

Those two things are tested completely differently in a lab. We had built for the first one.

In hindsight this attempt was avoidable. It would have taken a single question before we started: how will this bottle be shipped?

◆ 第二幕

On the second attempt we deepened the sealing groove in the neck, giving the gasket more room to deform under load. The direction was right — but beyond the seal itself, it moved other things.

The neck dimension had changed, so alignment at the filling station changed with it. We saw a slight film of product on the neck wall, and greater variation in fill weight.

It was in the internal review that we recognised something: a bottle dimension is never changed in isolation. One dimension touches at least three things — contact area with the formula, alignment accuracy at the filling station, and force distribution at capping. On the second attempt we reviewed the seal and nothing else.

By this point the atmosphere had tightened. The client's project manager wrote that our sample was "not professional".

That stung. We have been making bottles for over a decade.

But the real turning point came on a three-way call.

Alan was in that call. She runs client communication for us, working in Chinese, English and French. After both sides had said the same things for the third time, she did not rush to translate the technical explanation we had prepared. She asked the client a question first:

"When you say it must not leak — in what situation, exactly?"

The client answered: the product sits in a carton, the carton lies on its side on a warehouse pallet, and it travels from a European warehouse to stores over two or three days.

That was the moment everyone realised the two sides had been discussing two different things. We were discussing cap torque. The client was discussing static sealing while lying on its side.

Language was only the surface. The real problem was that the requirement had never been aligned — and we had not noticed that our idea of "sealed" and the client's idea of "sealed" were not the same idea.

◆ 第三幕

Before the third sample, Hany raised a question in the meeting that none of us had asked:

"We keep asking how to make the seal better. What if the problem is not the sealing structure, but the width of the contact face between the neck and the gasket?"

That single question turned the discussion around. We stopped asking "how do we make A better" and first established whether A was even the critical item.

The third design reworked the sealing structure at the neck, and changed the gasket to a material that returns more consistently after compression. We sent the sample. The client ran a static side-lay test.

It passed.

◆ 尾聲

Along with the sample we attached something: a table showing what had changed at each attempt, why, and which judgement had been wrong.

Their procurement director said something we still quote:

"You got it wrong the first time. But you knew where you went wrong."

The one thing we did right: we showed the reasoning, not just the conclusion. The client did not need us to say "this is better". He needed to know why — because he signs the order, and he carries the launch risk.

The order was placed, and we still work together today. But more important than the order were three things we changed afterwards:

First, align the scenario before the parameters. Our sample request form now has a mandatory field: shipping method and consumer use action.

Second, any dimension change needs a knock-on review. One dimension affects compatibility, alignment and load. Not just on the drawing — we run it on the filling line.

Third, replace the adjective with a scenario. When a client says "good seal", "premium feel" or "looks expensive", we now ask them to convert that word into a concrete situation: how the bottle will be shipped, and how the consumer will use it.

Three samples for one bottle is not a small cost. But if it stopped us walking the same detour again, then those three attempts were worth it.

❓ Three questions

How many sample rounds is normal? It depends on the bottle and the sealing requirement. With a standard stock bottle, one round is usually enough. Where a new mould is needed, or the seal has a specific use case such as side-laid shipping, two or three rounds is common. The important part is asking, each time, why the previous version failed.

How long does sampling take? Around 5–7 working days for a standard bottle. Tooling is additional. Ask our team for the actual schedule — we give real dates, not optimistic ones.

What if the sample keeps failing? We break "failing" down into measurable items instead of repeatedly changing the design. The story above is the example: on the first two attempts we were changing parameters, when what actually needed changing was our understanding of the word "sealed".