Beauty Device Sampling: What Each Sample Proves, and How Long It Takes

A sample is not a sample

Most disputes in a device project start with two people meaning different things by “sample”. There are three kinds, and they prove completely different things:

Type	What it is	What it proves	Typical lead time
Stock sample	A unit off the shelf, already in production	The platform exists and its build quality	3–7 days
Golden sample	A reference unit agreed in writing as the standard	The exact build you are buying	1–2 weeks after specification sign-off
Pre-production sample (T1, T2)	A unit from the actual tooling	That production matches the design	2–4 weeks per round after tooling

Confusing these three is the most common cause of “this is not what I ordered”. A stock sample is not a commitment to your build. A golden sample is.

What a stock sample can and cannot tell you

A stock sample answers exactly one question: is the general build quality acceptable for the price point?

It can tell you about surface finish, seam quality, button feel, weight balance, noise level, motor vibration, and whether the product looks like the photos.

It cannot tell you:

Whether your housing shape will work — it is a different housing
Whether your firmware will run — it is a different board
Whether the colour will match your brand — it is a stock colour
Whether the factory can hit your specification — that is what tooling tests

Treat a stock sample as a supplier evaluation, not a product approval. This is the cheapest and most skipped step, and skipping it is why some projects discover an unworkable supplier only after tooling.

The golden sample

A golden sample is a physical unit that both sides agree represents the accepted build. It should be:

Signed and dated by both sides, not just sent by email
Accompanied by a written specification — dimensions, materials, colour codes, output values, packaging contents
Stored in two places — one at the factory, one at your office
Photographed from fixed angles so later batches can be compared

Without this, “matching the sample” is an opinion, and opinions in a dispute belong to whoever holds the money.

The golden sample is also the reference for the defect threshold. If the agreed standard is the golden sample, specify how much deviation is allowed — colour tolerance, seam width, surface defects per unit.

Pre-production samples: T1, T2, and why there is always a T2

After tooling, the factory moulds the first parts and builds a unit. That is T1.

T1 almost always has problems, and they are usually not the dramatic ones. They are:

A parting line in a place the design did not anticipate
A draft angle too small, causing drag marks
A clip that clicks in the CAD model and not in steel
A button that needs more force than specified
An interference between two parts discovered at assembly, not in the software

Each fix means a tool adjustment, another build, and 1–3 weeks. Budget for two rounds. A project that plans for one is planning to be late.

Typical timeline, stage by stage

Stage	Duration	What blocks it
Quotation and specification	3–7 days	Your answers
Industrial design	1–3 weeks	Approval rounds
Mechanical design + first prototype	2–4 weeks	Part complexity
Prototype review and changes	1–3 weeks	How many changes
Tooling	25–35 days	Part count, finish, tool complexity
T1 build and review	1–2 weeks	Your inspection
Tool correction + T2	2–4 weeks	Severity of T1 findings
Certification (EMC etc.)	3–6 weeks	Laboratory queue, retests
Pilot run	2–3 weeks	Material lead times
Production	25–40 days	Component availability

From specification to first container, a new device is realistically 5–9 months. An ODM adaptation is 1–3 months. A private-label order is 2–4 weeks.

What actually slows a sample round down

In order of how often it happens:

Unclear specification. The factory builds to the gap, then rebuilds. Two weeks lost per round.
A design change after tooling. Every change is a tool adjustment; some need a new tool.
Component lead time. A specific motor, screen or sensor can be 4–8 weeks on its own.
Slow approval. A sample sitting on someone’s desk for ten days is ten days added to delivery.
Laboratory queue. EMC labs in peak season run 3–4 weeks before testing even begins.

Only one of these five is the factory’s fault. The other four are in the buyer’s control, which is why buyers who pre-write the specification get samples faster than buyers who negotiate hardest.

How to make a sample round productive

Write the specification before the sample is ordered. One page is enough: dimensions, key materials, output values, colour codes, packaging contents, target markets.
State what you will actually check. If you will measure output, say which instrument and which tolerance. If you will not measure it, do not list it.
Limit the sample to the questions you need answered. A sample trying to prove five things proves none of them well.
Inspect within 48 hours of arrival and send findings in one message, not over a week.
Ask what the factory sees as the weak point in the sample. A supplier who names it before you do is worth keeping.
Pay for samples. A free sample is a favour, and favours do not come with specifications.

A sample inspection list

What to check on arrival, in the order that catches the most problems fastest:

Build

Seams: even width, no flash, no shadow line where two parts meet
Surface: no sink marks on large flat areas, no flow lines near the gate
Texture: consistent across parts, and consistent with the golden sample
Fit: battery cover, tank, cap, accessory mounts — all should seat without force
Moving parts: buttons, sliding covers, rotating heads — count the cycles they survive

Function

Output measured, not observed — use the instrument written in the specification
Charge and discharge cycle: does the runtime match the claim on the box
Heat: surface temperature after ten minutes of continuous use
Noise and vibration: compared against the golden sample, not against expectations

Documents

Test reports matching this model
Manual: language, warnings, and whether the claimed functions match the device
Packaging contents: everything on the box is in the box
Labels: the rating plate matches the certification

Commercial

The model number on the sample matches the quotation
Accessories are the specified ones, not substitutes

Two rules make this list work. Write down what you checked, so the next round is comparable. And measure rather than judge wherever an instrument exists — “the light looks dimmer” is not a finding, “irradiance at 10 cm is 22% below the agreed value” is.

What samples cost

A paid sample of an existing platform typically runs 2–4× the unit price, and it usually excludes freight. Custom samples from new tooling are not sold at unit price at all — the sample is effectively the first output of a paid prototype round, so the cost sits inside the development budget.

Ask two things when you are quoted a sample price:

Is this a stock unit or a build to my specification?
Is the sample cost credited against the first order? (Many factories credit it; it is worth asking, and worth getting in writing.)

How we sample

We tell you which of the three kinds a quoted sample is, what it will and will not prove, and what we expect to fix in the next round. If a stock sample is enough to answer your question, we will say so rather than sell you a prototype.