The first production run of a foundation almost always matches the approved sample, because the plant made the sample and the first batch weeks apart from the same pigment lots, on the same kettle, with the same technician adjusting the colour. The second order is where base makeup colour consistency between production lots gets tested. By then the yellow iron oxide is from a new drum, the titanium dioxide may be a different grade lot, the golden sample has sat in a drawer for four months, and the person who signed off the first batch has moved to another line. The finished cream is inside any tolerance the plant thinks reasonable — and your customer, holding both bottles side by side, says they do not match.

You cannot prevent that argument with goodwill. You prevent it with a tolerance written into the order, a measurement method both sides accept, a retained reference that has not aged out, and a batch record that lets you trace which pigment lot went into which finished lot. This article covers each of those, in the order you should put them in front of a supplier.

Why shade drifts even when the formula does not change

A foundation shade is built from a small number of pigments — typically titanium dioxide plus red, yellow and black iron oxides — dispersed into a base. The formula card fixes their percentages. It does not fix their colour, because pigments are natural-origin or synthetic minerals with a permitted range of tinting strength and undertone from lot to lot. Yellow iron oxide is the usual culprit: its hue can shift visibly between drums that are both inside the raw-material supplier's specification.

On top of pigment variation there are process contributors that a good plant controls and an average one does not:

  • Dispersion quality. Pigment that is not fully wetted and de-agglomerated reads darker and duller than the same pigment properly dispersed. A change in mill time, mill type or the order of addition moves the shade without any change on the formula card.
  • Batch size. A 50 kg development batch and a 500 kg production batch do not shear pigment identically. Shade approved on the small batch often needs adjusting on the large one — that is normal, and it is why the production batch has a colour-adjust step at all.
  • Water loss and temperature. Emulsions concentrate slightly during long hot holds; a shade measured wet at 70 °C and a shade measured after cool-down are not the same colour.
  • Dry-down. Foundation shifts as the volatile phase evaporates on skin. A plant that checks colour wet against a wet standard is checking the wrong endpoint if your customer complaint is about the colour after application.

None of this means the plant did anything wrong. It means "make it the same as last time" is not a specification, and the second order is the point at which that becomes expensive.

The colour-adjust step: what actually happens in the kettle

This is the part of the process most buyers have never seen, and it is where tolerance is won or lost.

A production batch of base makeup is not compounded to 100% and then measured. The usual practice is to build the batch with the pigment charge held back slightly — the exact hold-back is the technician's judgement — then draw a sample, compare it to the retained standard, and add small quantities of pigment slurry or pre-dispersed colour paste to walk the batch onto shade. That adjustment is repeated until the sample is judged inside tolerance, and only then is the batch released to filling.

Three things follow from that:

  1. The adjustment step is the control point. If you want to audit how a plant controls shade, ask to see the colour-adjust record for a previous batch: how many additions, of what, judged how. A plant that can produce that record has a process. A plant that cannot is relying on one person's eye.
  2. Pre-dispersed colour pastes reduce drift. Plants that pre-disperse each pigment into a standardised paste and adjust with paste rather than dry powder get tighter batch-to-batch results, because the dispersion variable is removed from the adjustment step. Ask which the supplier uses.
  3. Time to shade is a cost. Every adjustment round is kettle time. A plant under schedule pressure has an incentive to accept the third round rather than run a fourth. A written tolerance takes that decision away from the schedule.

Yadun Cosmetics, a Baiyun District contract manufacturer with base makeup on its first product line and compounding, lab and QC listed among its plant functions, is the kind of plant where this step exists. What its colour-adjust procedure looks like, and whether it adjusts with paste or powder, is not published on its factory page — ask, and ask to see a record.

Writing a tolerance: visual panel versus instrumental ΔE

There are two ways to define "matches", and the contract should name one of them.

Method How it works Strength Weakness What to write
Visual panel Trained assessors compare drawdown or skin-swatch against the standard under a controlled light booth (typically D65 daylight, sometimes also store-type fluorescent) Catches undertone and finish differences an instrument can miss; cheap Subjective; depends on who is on the panel that day; hard to arbitrate Booth light source, number of assessors, pass rule (e.g. majority), drawdown method
Instrumental (spectrophotometer) Drawdown of standard and batch at fixed film thickness on a white card; instrument reports Lab* values; difference expressed as ΔE Objective, repeatable, can be done at both ends and compared by email Does not see finish or dry-down; a ΔE inside limit can still look "off" if the shift is all in one axis Instrument geometry (e.g. 45°/0°), illuminant/observer (D65/10°), ΔE formula (ΔE*ab or ΔE2000), drawdown thickness, limit
Both (recommended) Instrument as the gate, panel as the tiebreaker on undertone Objective release plus a human check Costs two steps Instrument limit plus panel sign-off on any batch within the outer half of the limit

On the number itself: many brand specifications for base makeup land somewhere around ΔE 1.0 to 1.5 against the standard, with tighter limits for the paler shades where small shifts are more visible. Treat that as a conversation opener, not a figure to copy in. The right limit depends on your shade range spacing — if two adjacent shades are only ΔE 2 apart, a tolerance of 1.5 lets a batch of one drift most of the way to the other — and on what your retail returns data actually shows customers noticing.

Two details that decide whether an instrumental clause is enforceable:

  • Name the ΔE formula. ΔE*ab (the 1976 formula) and ΔE2000 give different numbers for the same pair of colours, and the gap is largest exactly where foundation shades live. If the contract says "ΔE ≤ 1.0" without saying which, expect a dispute.
  • Fix the drawdown. Film thickness changes L* directly. Specify the applicator (bar or block), the gap, the card, and whether readings are taken wet or after a stated dry time. Without that, two labs measuring the same batch will disagree by more than the tolerance.

If the plant only offers visual assessment, that is workable for a pilot, but write the booth conditions and the panel size into the order anyway, and plan to add an instrument clause before the second run. Whether Yadun assesses colour visually, instrumentally or both, and to what limit, is one of the items you should request in writing before the first order — it is not on the factory page.

The golden sample problem: your reference is drifting too

A retained standard is a cosmetic emulsion in a jar. It oxidises, it loses volatiles through the seal, its pigments can settle, and after enough months it is no longer the colour it was on approval day. A plant that matches batch four against a two-year-old standard is matching to the wrong target — and the batch will look different from batch one even if the plant's process is perfect.

A retention clause worth proposing:

  1. Three sealed standards at approval, from the approved batch, split between the plant, the buyer, and a neutral third party or a second location. Filled to the top, opaque, cool storage.
  2. A stated retention period, after which a fresh standard is cut from an approved production batch and the old one is retired. Twelve months is a common starting point for emulsions; your stability data should confirm it.
  3. A re-certification step whenever a new standard replaces an old one: the new standard is measured against the original approval values (the Lab* numbers recorded at approval, not the ageing jar), and the drift is documented.
  4. A retained sample from every production lot, held for the shelf life of the product plus a margin, so a complaint on lot 7 can be checked against lot 7 and not against memory.

Point 3 is the one most contracts miss. The instrument values recorded on approval day are the only reference that does not age. Get them written into the spec sheet — the same document that carries the formula code, the pack and the label, following the discipline in this guide to writing a product spec sheet for a Chinese factory — and have both sides sign it. How long Yadun retains standards and lot samples is not published; propose the clause and see whether the plant already does something equivalent.

Pigment lot control and the batch record that makes it traceable

Tolerance and standards manage the output. Pigment control manages the input, and it is cheaper.

Ask the supplier how it approves an incoming pigment lot. The practices that separate a controlled plant from an uncontrolled one:

  • Incoming colour check per pigment lot — a drawdown of the new lot against a retained pigment standard, or at minimum a tint-strength comparison in a white base, before the drum goes into stock.
  • Lot segregation — a batch is made from one lot of each pigment wherever the batch size allows, rather than the tail of one drum and the head of the next.
  • Raw-material approval procedure — a written rule for what happens when a pigment lot is outside its expected range: reject, or accept with a documented adjustment to the formula.
  • Supplier change control — a commitment that a pigment or titanium dioxide grade is not substituted without notice. Titanium dioxide surface treatment in particular changes both shade and skin feel, and substitutions here are a frequent hidden cause of "the second lot feels different".

Then ask what arrives with the shipment. A useful batch record for base makeup contains, at minimum:

Item Why you need it
Finished-lot code, and the format it follows So the code on the bottle maps to a record
Compounding batch number(s) feeding the lot One filled lot may draw on more than one kettle batch
Pigment and TiO2 lot numbers used Lets a complaint be traced to an input
Colour-adjust record Additions made, judged how, by whom
Release colour reading or panel result vs standard The evidence the batch was inside tolerance when released
Fill date and release date For shelf-life and retention calculations

You will not always get all of it. Ask for it before the order, so the absence of any line is a negotiation rather than a surprise. The format of the batch record and lot code that Yadun supplies with shipments is not published; request a sample record from a previous product as part of qualification.

Building the clause set into the order

Shade control does not fit in one sentence on a purchase order. It fits in a short annex. A workable structure:

  1. Reference: approved standard identifier, approval date, recorded Lab* under stated instrument conditions.
  2. Method: instrument, geometry, illuminant/observer, ΔE formula, drawdown procedure; panel conditions if used.
  3. Limit: ΔE maximum per shade, and any tighter limit for named pale shades.
  4. Release: batch is not filled until inside limit; release reading recorded and supplied.
  5. Retention: number of standards, holders, retention period, re-certification rule, per-lot retained sample.
  6. Inputs: incoming pigment check, lot segregation, change-control notice for pigment and TiO2 grade.
  7. Records: batch record contents supplied with each shipment.
  8. Dispute: if buyer and plant readings disagree, which lab is the referee and who pays.
  9. Remedy: what happens to an out-of-tolerance lot — rework by adjustment where possible, replacement, or credit.

Point 8 is where the Baiyun cluster helps in a practical way: testing labs sit within the district, as Yadun's page notes, so a referee reading on a disputed lot can be turned round locally rather than shipped abroad. A pre-shipment inspection that includes a colour reading against the retained standard — whether by your own inspector or a third party — is the last cheap check before freight; the general framework is in this guide to product inspection in China before shipment.

If you developed the range with the same plant, the shade-spacing question from that stage feeds directly into the tolerance here — see the companion article on building a foundation shade range with a Chinese base-makeup manufacturer.

Common questions

The plant says "colour matches every time" and will not sign a ΔE limit. What does that tell me?

That it assesses visually and has not been asked for a number before. That is not disqualifying for a pilot, but write the visual conditions down, insist on the approval-day instrument reading from a local lab so you own the reference values, and revisit before the second order.

Does a 15–30 day production window include colour adjustment?

Yadun states 15–30 days production after sample approval. Whether that window assumes a batch that hits shade on the first adjustment round or leaves room for several is not published. Ask, and ask what happens to the schedule if a batch needs a rework.

My first order was 2,000 units. Will the second batch be made the same size?

Only if you order the same quantity. A larger second order means a larger kettle batch, which shears pigment differently and may sit at a different point in the colour-adjust process. Flag any change in batch size to the plant explicitly and expect an extra adjustment round on the first batch at the new size.

What to ask the supplier next

  1. How is shade judged at release — visual, instrument, or both — and what limit is applied today?
  2. Can you share a colour-adjust record and a release reading from a previous base-makeup batch?
  3. Do you adjust with pre-dispersed paste or dry pigment?
  4. How are incoming pigment and titanium dioxide lots checked, and will you notify grade changes?
  5. How many standards are retained, where, for how long, and how are they refreshed?
  6. What does the batch record and lot code supplied with a shipment contain? Ask for a sample.
  7. Which local lab would act as referee on a disputed reading?
  8. What is the remedy for an out-of-tolerance lot?

Yadun Cosmetics is a plant of the type these questions are aimed at: a Guangzhou contract manufacturer operating since 2009 with 100–500 staff, base makeup on its first product line, compounding, lab and QC teams on site, standard sampling in three days and custom in seven, and production quoted at 15–30 days after sample approval. Those are company-stated terms. The shade-control procedure behind them is not published, which is why the clause set above exists — read the factory profile for what is stated, and get the rest in writing before the second order, not after it.