Cosmetic Certificate of Analysis (COA): What Brands Should Specify Before the First Order
The short answer
A cosmetic certificate of analysis (COA) reports what one batch measured against limits that were agreed in advance. The approved finished-product specification is where those limits are agreed. If no specification is signed off before the first order, the supplier effectively decides which parameters the COA shows and what counts as a pass.
Before the first purchase order, a brand should approve a specification tied to a defined formula version. It should list each parameter, its test method and its acceptance limit. The brand and manufacturer should also agree which of those items are batch-release tests that the COA will report. Each batch COA should then cite the specification revision in effect for that batch and report actual results for the agreed items. A COA alone does not show that a product is safe, legally compliant or approved for any market. This article focuses on the buyer's document checks; our cosmetic manufacturing process guide covers the wider production sequence.
Four documents buyers often confuse
Swipe horizontally to read the full table on a narrow screen.
| Document | What it records | Scope and usual issuer | What it cannot show on its own |
|---|---|---|---|
| Approved finished-product specification | Parameters, test methods and acceptance limits, and which items are batch-release tests | One formula version, valid until revised under change control; drafted by either party, approved in writing by both | That any batch met it |
| Batch COA (finished product) | Batch-specific results for the agreed reported items against the applicable specification; release status or decision only if the parties include it | One batch or lot; issuer and approval role depend on the agreed quality arrangements | Product safety, regulatory compliance, market approval, or anything outside the items reported |
| Separate third-party laboratory report | Results and conditions for the tests commissioned from an outside laboratory, such as a preservative efficacy or compatibility study | A defined sample or study design; issued by the external laboratory for its stated scope | Performance of later production batches or tests outside the report scope |
| GMP certificate, audit statement or supplier declaration | An independent certificate or third-party audit statement records an external assessment against a stated scope and date; a supplier self-declaration states its own claim and is not equivalent independent verification | A facility or quality system, not a product; identify the actual issuer and evidence type | Whether any particular batch met its specification |
Approved finished-product specification
- What it records
- Parameters, test methods and acceptance limits, and which items are batch-release tests
- Scope and usual issuer
- One formula version, valid until revised under change control; drafted by either party, approved in writing by both
- What it cannot show on its own
- That any batch met it
Batch COA (finished product)
- What it records
- Batch-specific results for the agreed reported items against the applicable specification; release status or decision only if the parties include it
- Scope and usual issuer
- One batch or lot; issuer and approval role depend on the agreed quality arrangements
- What it cannot show on its own
- Product safety, regulatory compliance, market approval, or anything outside the items reported
Separate third-party laboratory report
- What it records
- Results and conditions for the tests commissioned from an outside laboratory, such as a preservative efficacy or compatibility study
- Scope and usual issuer
- A defined sample or study design; issued by the external laboratory for its stated scope
- What it cannot show on its own
- Performance of later production batches or tests outside the report scope
GMP certificate, audit statement or supplier declaration
- What it records
- An independent certificate or third-party audit statement records an external assessment against a stated scope and date; a supplier self-declaration states its own claim and is not equivalent independent verification
- Scope and usual issuer
- A facility or quality system, not a product; identify the actual issuer and evidence type
- What it cannot show on its own
- Whether any particular batch met its specification
A raw-material COA is a separate document. It comes from the ingredient supplier and describes one lot of one ingredient. A clean raw-material COA for a preservative tells you nothing about the pH or microbial status of the filled jar. File raw-material and finished-product COAs separately, and never accept one in place of the other. Our quality and testing page shows a redacted Rosa damascena flower water (rose hydrosol) raw-material batch COA; it is not a finished-product COA template.
Why the specification comes first
The supplier chooses the parameters. A COA can show appearance, colour and odour as "conforms", leave out pH or microbial results, and still look complete.
The limits can drift. If no specification was approved, nothing stops a later COA from printing a wider pH range than the first one.
The method is undefined. A viscosity figure means little without the instrument, spindle, speed and temperature.
The specification closes these gaps. It does not have to put every control on the COA. Some items, such as a stability study or a preservative efficacy test, are one-time or periodic studies kept in their own reports. Others may be controlled through raw-material or in-process records. Agree in writing which items are batch-release tests reported on each COA, and where the rest are recorded.
What the regulations say, and what they don't
A buyer needs to separate two questions: What evidence does the market require to support a safe, correctly presented product? and Which characteristics will the manufacturer report for this particular batch? The first question is broader than a COA. The second belongs in the approved specification and the parties' quality arrangements. None of the official sources below creates a universal finished-product COA form or an identical test list for every cosmetic. Market rules and standards editions should be checked again when a product is launched or an agreement is signed.
European Union: GMP, safety assessment and the product information file
Regulation (EC) No 1223/2009, consolidated text requires manufacture in accordance with good manufacturing practice (Article 8), a cosmetic product safety assessment before a product is placed on the market (Article 10), and a product information file (PIF) for the product (Article 11). Annex I describes the safety report's content, including the product's composition, physical and chemical characteristics and stability, microbiological quality, impurities and packaging information. A batch COA may contribute evidence about some measured characteristics, but it is not the safety report, the whole PIF or proof that the product has met every market requirement. The PIF retention rule—10 years after the last batch was placed on the market—concerns that file; do not casually copy it as the contractual retention period for every supplier record without considering the applicable obligation and agreement.
Claims require separate substantiation, not a COA checkbox. Where justified by the nature or effect of the cosmetic product, Article 11(2)(d) of Regulation (EC) No 1223/2009 calls for proof of the claimed effect in the product information file. Commission Regulation (EU) No 655/2013 sets common criteria for cosmetic claims, including adequate, verifiable evidence relevant to the product. A COA showing that one batch met an agreed pH or appearance range does not, by itself, substantiate a benefit claim. Nor does a raw-material claim automatically become an established claim for the finished formula. The brand should route proposed claims to its qualified regulatory and safety contacts and identify the actual evidence location in the project file.
The European Commission's SCCS Notes of Guidance, 12th revision, Appendix 6 discusses a certificate of analysis for a cosmetic ingredient. Its list of analytical and identity details can help a buyer understand an ingredient-lot record; it is not a legal finished-product batch COA template. This distinction matters when a supplier sends a polished preservative or extract COA in response to a request for the finished cream's batch evidence.
United States: safety substantiation is wider than a batch result
The FDA's MoCRA overview explains that the responsible person must ensure, and maintain records supporting, adequate safety substantiation. FDA also says the law and regulations do not require a specific test for every individual cosmetic product or ingredient. The FDA's cosmetics product-testing page explains that cosmetics generally do not need FDA premarket approval, with a separate exception for colour additives. Thus neither a supplier COA nor a line saying “FDA approved” can be treated as market approval of the cosmetic. Ask what safety records, labeling review and other market-specific work the responsible person will hold separately from the batch COA.
FDA's linked Draft Guidance for Industry: Cosmetic Good Manufacturing Practices is explicitly draft guidance, not a finished-product COA form or, by itself, a binding final GMP regulation. MoCRA also directs FDA to establish cosmetic GMP regulations. A supplier's use of a draft checklist or a GMP certificate does not prove that a particular batch met a product-specific specification. Check the current FDA rulemaking status when the project requires a regulatory conclusion; this guide does not substitute for that review.
Standards are references to verify, not automatic COA lines
The ISO 22716:2007 catalog entry identifies a cosmetics GMP guideline for production, control, storage and shipment. Refer to the current edition of ISO 17516 for cosmetics microbiological limits. The ISO 11930:2019 entry and its 2022 amendment concern evaluation of a cosmetic product's antimicrobial protection. The public catalog abstracts are not a license to copy numerical limits into every project or to claim every batch needs a preservative-efficacy study. If the parties cite a standard, write its edition, the purpose for which it is cited, the project-specific method or criterion, and the evidence record. Have qualified staff check the full applicable text and the product's current market requirements.
For Great Britain, the UK government cosmetics guidance discusses responsibilities including the responsible person, product information file, safety assessment and GMP for England, Scotland and Wales; Northern Ireland has a distinct regime. A GB project needs its own market review rather than copying an EU or US checklist. The commercial question remains the same: which batch fields does the brand require on each COA, and which separate legal or safety records must be maintained elsewhere? For separately commissioned tests, our cosmetic testing laboratory guide explains how to define a laboratory brief and report scope without treating a laboratory report as a batch COA.
If you are preparing a first-order brief, contact our team with the proposed product format, formula status, pack and target markets. Ask which specification and batch-record fields can be agreed for your actual project; a generic example here is not a factory commitment.
Before the first order: decisions to close
The signed specification should be an outcome of a decision, not a document that arrives attached to a quotation with no review. A brand can ask the manufacturer to draft it, or draft a version itself. In either case, the brand's product owner and quality contact should be able to name the exact formula and pack they are accepting, which characteristics will control batch release, and which studies sit in separate reports. The manufacturer should confirm that the chosen methods and acceptance criteria can actually be applied. The parties then record who approved the version and when it becomes effective for the first affected batch. This is a suggested procurement workflow, not a universal legal sequence or a claim about any particular factory.
A useful first-order decision packet contains four things: the buyer brief; the approved formula and primary-pack identifiers; the controlled specification revision; and a decision record for questions that do not belong inside the numerical table. That decision record matters because a vague phrase such as “COA provided” does not tell the brand whether it will see a signed PDF before shipment, with shipment or only on request. Nor does it tell quality staff whether shipment waits for the brand's review. Those are commercial and operational choices to settle in writing, not a default supplier promise. The first order can proceed with an open noncritical study only if the parties have consciously documented what remains open and what future activity is blocked. Conditional approval of a document is not evidence that a batch is released or ready for market sale.
A short decision record
Fictional buyer exercise; not a real manufacturer's record, quotation, service commitment or recommended specification. The codes and dates are invented.
Swipe horizontally to read the full table on a narrow screen.
| Decision | Recorded answer in this fictional project | Evidence or owner | Status before order |
|---|---|---|---|
| Formula and primary pack | Formula F-003 with pack PK-50-B; other development versions excluded | Brand product lead keeps the signed reference list | Closed |
| COA release scope | Appearance and two named numerical tests proposed; manufacturer to confirm method and limits | Brand quality lead and manufacturer quality lead | Open: specification cannot be signed yet |
| Separate study | Compatibility report to address F-003 with PK-50-B; applicability to be assessed after a pack change | Report owner to provide the study ID and scope | Open: launch decision depends on review |
| COA handoff | Brand requests a signed PDF before shipment and asks whether shipment waits for brand review | Buyer records written supplier response in supply terms | Open: no delivery time is assumed |
| Reference sample | Both parties identify the same approved sample and record storage and replacement responsibilities | Brand and manufacturer sample custodians | Closed after sample ID is agreed |
An “open” row is not automatically a stop for every commercial step. It is a stop for whichever decision the parties said it blocks. For the fictional release-scope row, there is no signed release specification while methods and limits are unresolved. For the fictional study row, the brand might continue packaging design, but should not infer that the missing report is complete. A useful decision record names the owner, the evidence still needed and the decision that cannot be made yet. Avoid leaving this logic buried in emails.
Reference samples and document timing
Numbers are not the whole specification. Appearance, colour and odour may be judged against a physical reference sample. The brand should ask which approved batch supplied that sample, whether both parties hold comparable retains, how it is stored, who replaces it when it ages and what happens if two retained samples no longer match. A COA line saying “appearance conforms” is weak if nobody can identify the standard. For a colour-sensitive item, the specification can also name lighting or instrument conditions rather than leaving the test to memory.
Before the purchase order, put each document on a handoff list: controlled specification, separate study reports, batch COA, batch or filling record extract where agreed, and any market-specific file supplied through a separate process. For each item, record the formula and pack revision it applies to, issuer, recipient, format, planned availability and the event it gates. A report promised “later” needs a named milestone. A batch COA delivery term needs a named trigger. The buyer should ask who can approve a corrected COA, how the new version supersedes the first, and whether a delayed record pauses shipping or merely document acceptance. The answer depends on the agreement and the product risk; this article does not assume the manufacturer has a fixed delivery schedule.
One final distinction protects the version history: the date printed on a new specification is not necessarily the answer to which batch it governs. The parties should specify the effective rule and first affected batch, including how work in progress is treated. Some arrangements use the manufacture date; others may tie effect to release or a named batch. A batch made before a change but released afterward is a predictable edge case. If the parties settle that edge case before ordering, procurement will not have to improvise when the COA arrives.
Format-Specific Specification Decisions
A finished-product specification is only useful if each line fits the physical product it describes. The same test name can mean different things for different formats. A pH reading on an emulsion, a viscosity reading on a rinse-off gel, or an appearance check on a stick after warm storage each raises a different question about method, sample state and acceptance criteria. Before the specification is approved, the brand and manufacturer should agree on these points for each format. Settling them in advance avoids arguments later about whether a result is valid.
The questions below are prompts for that conversation. They do not make up a required test panel, and they do not set limits. Which tests apply, and the range each one must meet, depend on the formula, the claims, the packaging, the target markets, and the manufacturer's own validated methods.
Emulsions (creams, lotions, emulsified serums)
Emulsions change after manufacture. Viscosity can drift while the structure settles, and the shear from pumping and filling can thin or thicken the product. For any rheology or appearance line, the buyer should know when and where the sample was taken.
Questions to ask:
- Is viscosity measured on bulk in the vessel, on bulk after a defined rest time, or on filled units? If filled units, how long after filling?
- Which instrument, spindle or geometry, speed, temperature and reading time does the method name? Will these be written on the COA or only in the method?
- For pH, is the emulsion type compatible with direct measurement? If the external phase is oil, is the product measured on a dilution, and at what ratio?
- Is "no separation" a release check on the batch, or does it come from a stability study kept in a separate report?
Water-based toners, mists, gels and serums
Water-rich formats raise microbiological questions, and their clarity and color are easy for consumers to see. Gels thickened with polymers can be very sensitive to pH and electrolytes, so a small pH shift can change viscosity.
Questions to ask:
- Which microbiological method is used, and is the sample taken from bulk, from filled units, or from both? Is there a defined hold time between manufacture and filling?
- How is clarity or color judged: visually against a retained reference, under defined lighting, or by instrument?
- For gels, is viscosity reported at a single speed, or does the method account for shear-thinning?
- For mists, which spray characteristics belong to the packaging interface rather than the bulk, and where are they recorded?
Anhydrous balms, sticks and oils
Many common water-phase tests do not carry over to anhydrous products. A pH reading on a pure oil has no clear meaning. For sticks and balms, structural questions may include thermal behavior and hardness; for oils, clarity, odor and oxidative change may matter more. The relevant controls depend on the actual formula and pack. The specification should state which lines do not apply instead of leaving them blank.
Questions to ask:
- Is a melting or drop point, hardness or penetration value, or another structural measure part of release? At what conditioning temperature and after how long?
- Is the sample a molded stick, a poured pan, or a portion of bulk melt? Results from these can differ.
- If the manufacturer applies reduced microbiological testing based on low water activity or composition, where is that rationale documented, and does it cover the current formula revision?
- Are blooming, sweating or rancid odor checked at release, or only in a stability study?
Powders (loose and pressed)
For powders, possible controls include fineness, color match, moisture and, for a pressed cake, mechanical strength. These depend heavily on how the sample is prepared.
Questions to ask:
- How is shade compared: as a drawdown or swatch, on what substrate, against which retained standard, and under what light source? Is the comparison visual or instrumental?
- For pressed products, is a drop or breakage test part of release? From what height, onto what surface, with how many units?
- Is fineness or particle size checked by sieve or another method, and on bulk or finished pans?
- Is moisture or loss on drying a release test? If so, under which conditions?
Rinse-off products (cleansers, shampoos, body washes)
Surfactant systems often get their viscosity from salt and pH. As a result, the viscosity range and the pH method are closely linked. Many rinse-off pH methods use a dilution, which has to be stated.
Questions to ask:
- Is pH measured neat or on a stated dilution, and is the same method used for the specification and the COA?
- Does the viscosity method define temperature tightly enough for a thickened surfactant system?
- Are low-temperature clarity, foam behavior or active-matter content release tests, characterization data, or not tested? Each answer changes where the record lives.
Final packaging interface
Some finished-product properties exist only once bulk meets the container. These include net content, closure integrity, pump output, the legibility and permanence of the batch code, and compatibility with the container material. They are often controlled through filling and packaging records rather than the laboratory COA.
Questions to ask:
- Is net content checked by weight or volume, against which declared quantity, at what sampling frequency, and where is the result recorded?
- Are closure torque, leak checks or pump-dose checks part of the finished-product specification, a separate packaging specification, or in-process controls?
- Does the batch code printed on the unit match the lot on the COA exactly, including any filling-line suffix?
- Which report covers package compatibility, and which formula and component revisions did it test?
Practical Interpretation
The questions above give a buyer three ways to read a draft specification.
- No line should be silently empty. If a property is not tested, the specification should say "not applicable" and give a short reason, or name the separate document that controls it. A blank invites someone to assume it was tested.
- The sample state is part of the limit. A viscosity range set on rested bulk cannot be compared directly with a reading taken from filled units the next morning. If the specification and the COA name different sample states, the two values are measuring different things.
- Sort each property into release, periodic, or characterization. Release lines are tested on every batch and usually appear on the COA. Periodic tests run on a defined schedule. Characterization data, such as a stability or compatibility study, may be done once per formula or package revision. Agreeing on this sorting tells the buyer in advance which fields to expect on a batch COA and which to look for elsewhere. Not every specification field belongs on every COA.
Setting a first-order document scope?
Share your product format, target market, packaging and draft specification. Ask which batch-release records and delivery timing to agree before ordering; methods and limits depend on the project.
Ask About the COA Scope →Crosswalk: Approved Specification to Batch COA
A batch COA issued under the agreed quality system reports that one identified batch was assessed against named criteria with stated results. It does not establish product safety, regulatory compliance or market approval by itself. Those depend on other documents and responsibilities. The COA does become useful when each of its fields can be traced back to the approved specification revision.
Field-by-Field Crosswalk
Swipe horizontally to read the full table on a narrow screen.
| Field | Approved specification controls | COA and buyer comparison |
|---|---|---|
| Product and formula | Product code, formula revision and pack configuration | Confirm the COA identifies the same approved version; a marketing name alone is not enough. |
| Effective specification revision | Revision, approval history and first affected batch or other agreed effective rule | Compare the cited revision with the one applicable to this batch under that written rule; do not assume the latest revision applies backward. |
| Batch or lot identity | The mapping convention for bulk and, where relevant, filling lots | Match the COA lot to the unit/carton codes and any shipment record the agreement says will carry a lot identifier. Ask for a written bulk-to-fill mapping when codes differ. |
| Sampling state | Bulk, filled pack or both, with timing where the result depends on it | Check that the COA result came from the required state. A value inside the numerical range from the wrong state is not a match. |
| Method and conditions | Controlled method ID and revision; critical conditions such as dilution or temperature | Check the cited method revision. The COA may cite the controlled method instead of reprinting every condition; investigate any changed condition. |
| Limit, unit and precision | Approved criterion, unit and reporting rule | Compare the COA limit with the applicable revision. Equivalent units may be converted and documented; do not treat an unexplained unit change as a match. |
| Actual result | No result; it states what will be measured or observed | Request the measured value for numeric items when the parties agreed it would be reported. For descriptive items, trace “conforms” to a named reference. |
| Conformity and release | Agreed decision rule, release role and whether release is recorded on the COA or separately | Check reported conformity against the result. Find the release status, approver and date in the agreed record, which may be separate from the COA, and reconcile its batch identity. |
| Separate study | Report type, identifier, scope and trigger for review or repetition | A COA reference is not a new batch test. Check whether the study covers the current formula and pack, or whether a documented impact assessment supports bridging. |
Handling Fields That Need Judgment
Revision control. Ask the manufacturer to print the specification revision on every COA. If a specification changes during production or release activity, the buyer needs the agreed effective rule and first affected batch so each COA can be checked against its applicable revision.
Lot identity across bulk and fill. One bulk batch may be filled over several days or on several lines. If the specification defines tests on filled units, the COA should identify the filling lot the samples came from. That lot should also be the one printed on the unit.
Method and conditions. A method ID is only traceable if the buyer can obtain a summary of what that method specifies. The COA does not need to repeat the whole method. It does need to state conditions that change the number, such as temperature, dilution or spindle and speed, unless the cited method revision fixes them.
Precision and rounding. If the limit is "5.0–6.0," a reported result of "6" is ambiguous. Agree on reporting precision in the specification so results and limits use the same number of decimal places.
Studies kept in separate reports. Stability, preservative efficacy, compatibility and similar studies are usually done on representative batches, not on each production batch. If the COA mentions them, the entry should be a reference: report ID, the formula and package revisions studied, and the date. It should not read as a pass result for the batch. The buyer then checks whether the report directly covers the current formula and pack, or whether a documented impact assessment supports continued use after a change. That linkage belongs in the specification, so the COA does not have to carry it.
Fictional Mini-Example
Teaching example only. Every product name, code, number, limit, method and document below is invented to show how the comparison works. None of it is a recommended limit, a real supplier document, or a legal or regulatory template.
The example product is "Example Hydrating Cream," an invented oil-in-water emulsion. The invented specification is SPEC-EX-01 Rev C, approved 2026-03-01 and effective from fictional batch EX-B0001 under a written first-affected-batch rule. The invented COA is for bulk batch EX-B0001 and filling lot EX-B0001-F1.
Swipe horizontally to read the full table on a narrow screen.
| Field | Specification SPEC-EX-01 Rev C (invented) | Batch COA EX-B0001 (invented) | Buyer check |
|---|---|---|---|
| Product and formula identity | Example Hydrating Cream; formula FX-100 Rev 2 | Example Hydrating Cream; FX-100 Rev 2 | Match |
| Specification revision | Rev C, first affected batch EX-B0001 | Released against Rev C | Match under the fictional agreed batch rule |
| Batch/lot | Bulk lot plus filling suffix | Bulk EX-B0001; filling lot EX-B0001-F1 | Match; confirm unit print reads EX-B0001-F1 |
| Appearance | White cream, conforms to retained standard RS-EX-01 | Conforms to RS-EX-01 | Match |
| pH: method and sample | TM-PH-01 Rev 2, neat, 25 °C, bulk | TM-PH-01 Rev 2, neat, 25 °C, bulk | Match |
| pH: limit and result | 5.0–6.0 | 5.6; conforms | Within limit; same precision |
| Viscosity: method | TM-VIS-03 Rev 4; rotational viscometer, spindle S-5, 10 rpm, 25 °C, 60 s reading | TM-VIS-03 Rev 4; spindle S-5, 10 rpm, 25 °C, 60 s | Match |
| Viscosity: sampling point | Filled units, 24 h after filling | Bulk, before filling | Mismatch. Query before acceptance. |
| Viscosity: limit and result | 18,000–26,000 mPa·s | 21,400 mPa·s; conforms | Value is inside the range, but it was taken from the wrong sample state |
| Net content | Controlled in filling record FR-EX-01; not on COA | Not listed | Expected; request the filling record extract |
| Stability | Reference report STAB-EX-01 (FX-100 Rev 2, jar component J-2) | "See STAB-EX-01" | Reference only; confirm jar component is still J-2 |
| Overall decision and approver | Release by QC Manager role | Conforms; QC Manager; released 2026-04-14; tests dated 2026-04-10 to 2026-04-13 | Supplier marked the batch released after testing, but the buyer must pause acceptance because the viscosity sampling point does not match |
Reminder: the numbers and documents above are invented for teaching. The lesson is the viscosity row. The reported value sits inside the range, and a quick check would accept it. But the specification set that range for filled units 24 hours after filling, and the COA reports a bulk reading taken before filling. The buyer should not decide whether the batch conforms. The buyer should pause document acceptance and ask for the required filled-unit result. If that result cannot be supplied, a mere explanation is insufficient: manufacturer quality staff must document the deviation and impact assessment, and the brand's authorised quality contact must decide under the agreed terms whether any alternative evidence can be accepted.
A clean crosswalk like this one shows only that the COA matches the approved specification line by line. Whether the specification is adequate for the product's safety, claims and markets is a separate question, answered through other documents and by the people responsible for them.
Receipt Audit for One Batch COA
Work through the steps in order. Each step relies on the one before it. For example, it is pointless to compare results against limits until you know which specification revision applies and which batch the document describes.
Procurement can usually complete Steps 1 to 4 and the comparison in Step 7. Steps 5, 6 and 8, and every exception, should go to the buyer's quality staff or a nominated quality contact.
Step 1: Confirm the document type, file and version
- Document type. Confirm the file is a batch-specific COA for the finished product. Specification sheets, technical data sheets, raw-material COAs, stability summaries and "typical values" sheets are different documents.
- Issuer. Record the issuing entity and site. They should match the manufacturer or laboratory named in your supply agreement or approved spec.
- Document control. Record the COA number, issue date and page numbering ("page 2 of 3"). Note whether the document is an original or a reissue. A reissued COA should reference the document it replaces and give a reason.
- File integrity. Compare the layout, signature block and numbering with earlier COAs from the same issuer. Signs of editing, such as mismatched fonts or a result field that looks overwritten, are reasons to ask questions. They do not prove falsification. Ask the issuer to confirm the document through an agreed channel. Do not decide the outcome yourself.
Step 2: Check the product identity and formula revision
Check that the product name, internal product code and formula revision match the purchase order and approved specification. Where artwork carries a relevant product or batch identifier, reconcile it too; artwork does not necessarily print a formula revision. If the COA names a formula revision that your records do not show as approved, stop here and escalate. Every later step depends on knowing which product was made.
Step 3: Identify the effective specification revision
The COA should cite the specification number and revision used for release. Look up which revision was effective for this batch. Use the first-affected-batch number or effective date recorded in the change history, as described in the change-control section below. Then compare that revision with the one cited on the COA. If they differ, the limits printed on the COA may be the wrong limits, even when every result passes them.
Step 4: Trace the batch identity across pack and shipping records
Compare the batch or lot code on the COA with the unit and shipper codes. Then check the packing list, invoice, retained-sample label or other record only where the parties agreed that record would carry a lot identifier. Not every commercial document does.
Codes are often formatted differently. For example, one bulk batch may be filled into several fill lots, or the carton may carry a date code instead of the batch number. In that case, ask for a written mapping that links each fill lot to the bulk batch and to the COA. Do not accept a mapping that someone explains verbally but has not recorded.
Step 5: Check the sampling state
Confirm what was tested and when:
- bulk before filling, filled finished product, or both
- where the sample was taken (for example, the start, middle or end of the fill)
- the sampling date compared with the manufacture and fill dates
- whether microbiological samples came from the bulk or from filled units
If the approved spec requires testing in the final pack and the COA reports bulk results only, record that as a gap, even if every value passes.
Step 6: Check the method and test conditions
Each test item should reference its controlled method and revision; the COA can cite that method instead of reprinting every condition. The buyer needs a way to verify conditions that change the result. Typical conditions are:
- viscosity: temperature, instrument, spindle and speed
- pH: whether it was measured neat or diluted, and the dilution ratio
- appearance and odor: comparison against a named reference standard
A pH result measured on a 10% dilution cannot be compared with a limit set for neat measurement. Quality staff should confirm that the method on the COA is the one named in the effective spec revision.
Step 7: Check the units and limits
Compare each limit printed on the COA with the effective specification revision, checking both meaning and units. A mathematically equivalent unit conversion can be acceptable after it is verified and documented; an unexplained change is not. For dynamic viscosity, cP and mPa·s are equivalent units. CFU/g and CFU/mL use different sample bases and cannot be directly interchanged without a verified product-specific conversion and comparable sampling and methods. Check:
- units (including whether a microbial result is per gram or per milliliter)
- range end points and whether they are inclusive
- reporting precision and rounding
A COA that prints its own limits is a convenience. The approved spec is the reference.
Step 8: Read the actual results
The brand's quality reviewer should read the reported numeric values and compare them with the effective specification limits; procurement can flag missing values but should escalate interpretation. For a descriptive item, a "conforms" result needs a named standard or method. For a numeric item, the buyer should agree in advance that the COA will show the actual value; if it shows only "pass", ask for the value and decide against the agreed document requirement rather than claiming a universal legal prohibition.
The quality reviewer should also compare the values with earlier batches tested by the same method. A result that is within limits but far from the usual range is worth asking about. It is not a failure.
Step 9: Check sign-off and release status
Check the issuer and any test or review sign-off that the parties agreed the COA would carry. Then check the agreed release record, which may be a separate controlled document, for the batch identity, release status, authorised approver and date. Reconcile it with the COA and the shipment record. A COA need not itself state the release decision or name the person who released the batch.
Compare the documented release date with the ship date. A batch shipped before its documented release, or a conditional status without a stated condition in the relevant record, needs a written explanation and quality review.
Administrative mismatch or out-of-limit result
Sort each finding into one of two categories before you escalate.
- Document omission or administrative mismatch. The batch may conform, but the record is incomplete or wrong. Examples are a missing page, a missing method reference, "pass" in place of a value, a batch code typo that the mapping resolves, or a missing signature. The fix is a corrected or reissued document that references the original.
- Out-of-limit result. A reported value falls outside the limit in the effective spec revision. The record may be complete. The value may reflect a product, sampling or analytical problem; the cause is not established until the investigation is complete.
Some administrative findings can hide a result problem. If the COA cites the wrong spec revision, the results must be checked again against the correct limits. If a numeric item says only "pass", there is no value to check.
When to hold. Hold the affected batch pending a written investigation when any of these is true:
- a reported result is outside the effective limits
- the batch identity cannot be traced across the COA, the packs and the shipping records
- a test required for release under the effective spec was not reported
- a release test used a method that was not approved
- a passing result replaced an earlier failing result and no investigation record exists
"Hold" here means the affected units are not used, sold or distributed until the buyer's quality function decides. It is an internal business control. Any regulatory notification or market action must follow the buyer's own procedures and the applicable requirements for each market. Those need to be verified separately for each market and are not covered here.
Exception Triage Table
Swipe horizontally to read the full table on a narrow screen.
| Exception | What procurement can check | Evidence to request | What not to assume |
|---|---|---|---|
| Wrong document type (spec sheet, TDS, raw-material COA or typical values sent in place of a batch COA) | Title, presence of a batch number, results compared with "typical" ranges, issuer | The batch-specific finished-product COA for the stated batch | That typical values describe this batch, or that a raw-material COA covers the finished product |
| Wrong batch | Batch code on the COA against labels, cartons and any shipment document agreed to carry a lot identifier; fill-lot mapping | The correct COA, or a signed fill-lot to bulk-batch mapping | That the received batch matches the batch on the COA because the product name matches |
| Missing item (a release test absent or reported as "pass" without a value) | The effective spec's release item list against the COA lines | The missing value, the raw data reference, or a written statement that the test was not performed and why | That the test was done and simply left off, or that the item can be waived without a recorded decision |
| Wrong spec revision cited | The revision cited on the COA against the change history and the batch's effective revision | A COA reissued against the correct revision, and confirmation of which revision was used for release | That the results conform because they pass the cited revision's limits; check them against the correct limits |
| Result outside limit | The reported value against the effective limit, units and rounding | The investigation record: lab error assessment, root cause, impact on other batches, proposed disposition | That a value close to the limit is acceptable, or that a supplier's "acceptable" note overrides the approved limit |
| Unapproved method change | The method reference and revision on the COA against the spec; changed conditions (temperature, dilution, spindle) | The change record, method comparison or equivalence data, and whether the buyer approved the change | That a new method gives comparable numbers, or that results remain valid against the old limits |
| Passing retest after an initial out-of-limit result | Whether the COA shows the original result, retest count and dates; whether the reported value is the retest only | The investigation for the original result, including the reason for invalidating it, the retest plan, and all results from the original sample and any resamples | That the passing retest cancels the first result, or that averaging or picking the best result was justified without a documented, assignable cause |
Note on retests. A retest can help an investigation, but a passing retest does not automatically erase the original out-of-limit result. Quality staff should review the original data, sampling and method, the reason for retesting, all retest results and the documented disposition. A confirmed analytical or sampling error may change how the first result is interpreted; without a supported conclusion, procurement should not accept a COA that reports only the passing value while hiding the earlier result.
Change Control Across Batches
A specification revision records criteria that were agreed at a point in time. Batches are made continuously. Change control connects the two so that each batch is judged against the revision applicable under the parties' written effective rule, which may refer to manufacture, release or a named first affected batch.
Types of change and what they usually involve
The table describes what a buyer might reasonably ask for in each case. The actual notification and approval obligations are set by the supply agreement and the applicable rules for each market, which need verification.
Swipe horizontally to read the full table on a narrow screen.
| Change | Typical spec effect | Impact assessment questions | Linked records to request |
|---|---|---|---|
| Routine new lot of an already approved raw material, same supplier and grade | Usually none | Does the lot meet the manufacturer's raw-material spec? | Normally kept in the manufacturer's records; ask only if the agreement says so or a problem is traced to it |
| Raw-material supplier or grade change | May need a new spec revision; a new formula revision if the formula records the material code | Same INCI name does not mean equivalent. Consider impurities, particle size, carrier, preservative compatibility, color or odor shift | Change record, new raw-material spec, compatibility or stability data, safety assessment review where applicable |
| Formula change | New formula revision; usually a new spec revision | Claims, labeling and ingredient list; safety assessment; stability; packaging compatibility | Formula revision, spec revision, label and artwork version, stability plan or data |
| Test method change | New spec revision citing the new method | Can old and new results be compared? Are the limits still meaningful? | Method revision, comparison or equivalence data, effective batch |
| Acceptance-criteria change | New spec revision | Tightened: what triggered it, and are older batches affected? Widened: what data supports the wider range? | Justification, supporting batch or stability data, approvals |
| Packaging change | Revise the spec if the approved pack description, pack-dependent method or criterion changes | Assess primary and secondary components for the actual product and claims; do not assume any change is immaterial | Component revision, impact assessment and any needed compatibility evidence |
| Process change (equipment, batch size, order of addition, temperatures, site) | May or may not change the spec | Can existing tests detect the effects of the change? | Change record, process validation or comparison batch data, site details if the site changes |
Effective dates and the change record
Define the effective point in writing, preferably naming the first affected batch and the relevant event. A rule might use manufacture, release or another agreed point. A date alone leaves unclear status for batches made before a change but tested, released or shipped after it.
A usable change record contains:
- change number, description and reason
- formula, spec, method and packaging revisions before and after
- first affected batch and effective date
- impact assessment, including whether batches already released are affected
- linked supporting records
- approvals from the parties named in the agreement
Comparing old and new batches
Keep a batch register. Each row should list the batch, manufacture date, formula revision, spec revision, method revisions and packaging component revisions, plus any change references. Judge each batch's COA against its own row. When you look at trends, mark the point where a method changed. Do not treat results from different methods as one directly comparable trend unless the methods have been assessed and their equivalence or conversion is documented.
Fictional buyer exercise; not a real manufacturer's document or recommended limit: revision 03 uses pH 5.0–7.0. Change record CR-12 tightens the fictional range to 5.5–6.5 in revision 04, with both parties approving an explicit first-affected-batch rule starting at A5. Batch A3 reported 6.9 and was released under revision 03; its prior release is not automatically undone by the later numeric change, though any new safety signal still needs an impact assessment. Batch A5 reports 6.8 and is outside revision 04, even though that value would have fitted revision 03.
Why a later revision does not automatically invalidate an older batch. A previously released batch was judged against the revision that applied under the agreed effective rule for that batch. Applying every later revision backward would turn each tightening or method improvement into a reason to question all previous stock, and the release record would lose its meaning. Separately, if a change was triggered by a quality or safety signal, the impact assessment must ask whether older batches carry that risk. That question is answered with evidence about those batches. It is not answered by re-applying the new limits.
Why an old revision cannot be used after a change takes effect. After the agreed effective point, the approved product, method or criterion may be different. The old limits may not apply to the new formula, may not detect what the change was meant to control, or may accept values the parties have agreed to reject. A COA that cites an old revision for a post-change batch may reflect an outdated template, a method or process transition, an unimplemented change, or another documentation problem. Ask which one, and treat the batch as unconfirmed until the answer is documented.
Case A: First Order of a Water-Based Toner
Fictional buyer exercise; not a real manufacturer's record or recommended limits. All product names, numbers, ranges and correspondence below are invented to show the buyer's reasoning.
Situation
A fictional skincare brand is considering its first order for a fictional 150 mL water-based toner. The formula contains a humectant blend, a botanical extract and a preservative system. The manufacturer has shared a draft finished-product specification, "TNR-01 Rev 01". Before approving it, the buyer wants to settle which controls must appear on every batch COA and which are better supported by separate reports.
Step 1: Sort each specification item by where its evidence belongs
The buyer goes through the draft line by line. For each item, the question is: does this property vary from batch to batch in a way a release test would catch? Or is it a property of the formula and pack that a study supports once and change control then protects?
Swipe horizontally to read the full table on a narrow screen.
| Specification item (fictional) | Draft value | Where the buyer expects evidence | Reasoning |
|---|---|---|---|
| Appearance | Clear, colorless to pale yellow liquid | Batch COA | Can drift with extract lots or process errors |
| Odor | Characteristic, matches retained standard | Batch COA | Same as above; needs a named reference standard |
| pH (neat, stated temperature) | 5.0–6.0 | Batch COA, numeric | Affects preservative performance and skin feel |
| Specific gravity | 1.000–1.020 | Batch COA, numeric | Quick check for dosing or water-phase errors |
| Microbiological quality | Criterion and frequency to be agreed for this fictional formula | Proposed batch COA if the project's assessment supports batch testing; otherwise document the chosen control | Water-rich products warrant a microbiological assessment, but there is no universal batch panel set by this teaching example |
| Specified organisms | Organisms, method and decision rule to be agreed | Evidence location to be agreed | “Absent” has meaning only with a named sample and method; this fictional row is not a standard test list |
| Fill volume | Nominal 150 mL, agreed tolerance | Batch COA or separate in-process record (to be agreed) | Often a filling-line control, not a lab test |
| Preservative efficacy | Meets agreed criteria | Separate challenge-test report on the approved formula and pack | A formula-level study that is not repeated per batch |
| Stability and packaging compatibility | Agreed conditions and duration | Separate stability report | Supports shelf life; one batch cannot show it |
| Specific impurities, if relevant | Defined only after reviewing ingredients and market requirements | Raw-material records or another agreed report | Selection depends on the actual formula and verified target-market requirements |
Two rules come out of this table. First, the specification should name the evidence location for each control: a batch COA, a separate report, an in-process record or a raw-material control. Second, a separate report should be checked against the ordered formula and pack. If either changed, a written impact assessment may support continued use or identify a need for new work; applicability should not be assumed.
Step 2: Sample exchange
Buyer to manufacturer (fictional):
Thank you for TNR-01 Rev 01. Before we approve it, please confirm:
- Which items will appear on each batch COA with numeric results, and which will be reported as conform/not conform?
- The method reference and test temperature for pH, and the microbial method reference.
- The challenge test and stability report IDs that apply to this exact formula code and the 150 mL PET bottle with pump. If none exist yet, please tell us what is planned and the expected scope (to be agreed).
- Whether fill volume is recorded on the COA or in a separate filling record we can request.
Manufacturer reply (fictional):
pH, specific gravity and microbial counts will be numeric. Appearance and odor are reported as conforms. For preservation, we can attach the supplier COA for the preservative blend.
Buyer follow-up (fictional):
Thanks. The preservative supplier's COA supports your incoming raw-material check. It does not show how the finished toner performs. For approval, we need the finished-product challenge test report for this formula and pack, or an agreed plan for one. Please also add the retained-standard reference to the appearance and odor lines.
Step 3: Keep raw-material COAs in their place
The buyer files raw-material COAs as supporting traceability. They are never treated as finished-product evidence. A basic identity check helps: does the document name the finished-product batch and formula, or only an incoming ingredient lot? An ingredient-lot record cannot replace evidence about the finished product or a study on the finished formulation.
Step 4: Approval decision
Swipe horizontally to read the full table on a narrow screen.
| Condition | Buyer action |
|---|---|
| All items have a named evidence source, methods named, separate reports match formula and pack | Approve Rev 01 in writing, citing the revision number |
| Separate reports are planned but incomplete | Record the open studies and a written gate for each. Development or a commercial order may proceed only under a documented risk decision; document approval is not batch release or permission to sell. |
| Manufacturer cannot name a method or evidence source for an item | Do not approve; ask for a revision or remove the item by mutual agreement |
Case B: Second Batch of an Emulsion with a Revision Mismatch
Fictional buyer exercise; not a real manufacturer's record or recommended limits. No real factory outcome is implied; the branches show possible paths, not a known result.
Situation
A brand receives the COA for the second production batch of a 50 g face cream (an oil-in-water emulsion). After batch 1, both parties approved "CRM-02 Rev 03" under a fictional signed change record that names batch CRM-B02 as the first affected batch. Rev 03 changed the viscosity method to a different spindle and speed and narrowed the pH range. The batch 2 COA has two problems:
- Its header cites Rev 02.
- It reports viscosity as 48,000 cP but does not clearly identify the method used. The fictional Rev 03 range is 20,000–40,000 cP under its named method; fictional Rev 02 used 35,000–55,000 cP under a different method. The number is a potential exception, but the two ranges cannot be compared across methods until the method and raw data are checked.
Every other result falls inside both revisions.
Step 1: Put the documents on hold
The buyer marks the COA "under review" and does not sign off on the batch. This pauses document acceptance. The affected units should not be used, sold or distributed while the brand's quality contact decides. Whether and how the physical goods are segregated follows the brand's own receiving procedure. The buyer records:
- the date the COA was received and who received it;
- the revision cited on the COA against the revision in the approved specification file;
- the specific result in question, copied exactly as printed, including units and method notes.
Step 2: Ask focused questions
Buyer to manufacturer (fictional):
Batch CRM-B02 COA cites CRM-02 Rev 02. Our signed change record makes Rev 03 effective from CRM-B02. Please clarify:
- Which method (spindle, speed, temperature) produced the 48,000 cP result?
- Was this batch made and released against Rev 03?
- If the result is against Rev 03, has an out-of-specification investigation been opened? What is its reference?
- Please confirm that the batch number, manufacture date and test date on the COA match the batch record and the unit labels.
Step 3: Check traceability on the buyer's side
While waiting for the reply, the buyer checks what can be verified without the manufacturer:
Swipe horizontally to read the full table on a narrow screen.
| Fictional buyer-side check | What this exercise shows | Next action |
|---|---|---|
| COA lot CRM-B02, unit code CRM-B02, carton code CRM-B02 | The three visible codes match | Keep the code images with the receipt record; ask for a bulk-to-fill mapping if the manufacturer uses a second internal code. |
| Unit artwork identifies the approved cream SKU | The product identifier matches; artwork does not print a formula revision | Check formula revision against the signed specification and manufacturer record, not the artwork alone. |
| COA test date 2026-04-11 and fictional manufacture date 2026-04-09 | The sequence is plausible | Ask for the underlying records if dates on the issued COA or batch record disagree. |
| COA cites Rev 02; signed change record says Rev 03 starts at CRM-B02 | Revision mismatch remains unresolved | Hold document acceptance and seek the manufacturer quality investigation. |
| COA omits the viscosity method revision | A value alone cannot be compared to the two method-specific ranges | Request the original measurement record, controlled method and QA disposition. |
Step 4: Possible investigation branches
The buyer does not decide what caused the problem. That conclusion comes from the manufacturer's investigation records. What the buyer does is define, in advance, what evidence each explanation needs.
Swipe horizontally to read the full table on a narrow screen.
| Manufacturer's explanation | Evidence the buyer asks for | Buyer's possible response |
|---|---|---|
| Wrong COA template or transcription error is claimed | Original measurement record, actual method and calculation, manufacturer QA investigation, correction history and impact on release | Review the written QA disposition first. A corrected COA alone does not establish that the batch met Rev 03. |
| The old Rev 02 method was used | Deviation record, method comparison, original result, any justified retest plan and complete results, QA impact assessment | A passing retest does not erase the first result. The brand quality owner decides only after the full record and effective revision are resolved. |
| Result is confirmed outside the applicable Rev 03 criterion | Manufacturer QA investigation, cause, other affected batches, product-risk assessment and proposed disposition | Maintain hold; accept, reject or pursue another disposition only through the brand's authorised quality process and applicable rules. |
| Manufacturer believes Rev 02 still applied | Both signed approval histories and first-affected-batch rule | Establish the valid revision before judging results; record and correct the change-control breakdown. |
Step 5: Close the review
Swipe horizontally to read the full table on a narrow screen.
| Final status | Required record |
|---|---|
| Accepted after investigation | COA citing the documented effective revision (Rev 03 if confirmed), original records, manufacturer QA disposition, brand reviewer and date. |
| Accepted with deviation | Original records, manufacturer QA investigation and disposition, written deviation and impact assessment, batch-specific scope, authorised brand approver, rationale and date. A retest or corrected COA alone does not close the exception. |
| Rejected | Written notice citing the specification item, revision and result. Disposition handled as agreed in the commercial terms |
| Unresolved | Escalate under the agreed exception route (see below). Keep the hold in place |
Whatever the outcome, the buyer asks one follow-up question: how will the manufacturer make sure future COAs are generated from the current approved revision?
Questions to Settle Before the Purchase Order
Use this list to agree terms for a quality agreement or a PO appendix. Our cosmetic manufacturing agreement guide covers the wider contract structure; here the focus is COA handoff and batch-document review. Treat every answer as something to be agreed and written down. Nothing here is a standard term, and none of it is legal wording. For regulatory duties in the target market, check current official rules and obtain qualified advice where needed; a purchase-order checklist is not a regulatory assessment.
COA supply
- At what point will the batch COA be supplied: before shipment, with the shipment, or after release? Is shipment conditional on the brand receiving it?
- Who receives it: named roles, a shared quality inbox, or both?
- In what format: signed PDF, a controlled template, a fixed file name convention?
- Will the COA cite the specification revision, the method references and the units for each numeric result?
Release and review responsibilities
- Who releases the batch at the manufacturer, and what does that release confirm?
- Does the brand carry out its own document review before accepting the goods? Within what working period after receipt (to be agreed)?
- Who signs corrected COAs, and how are corrections versioned?
Raw data and separate studies
- Can the brand request raw data excerpts (instrument printouts, microbial plate records) for a specific batch? Under what conditions?
- Which separate reports (challenge test, stability, compatibility, third-party tests) will be shared? Will they come as full reports or summaries?
- Will the brand be told when a separate study no longer covers the current formula or pack?
Retention and confidentiality
- How long will batch records, COAs and retained samples be kept under the applicable rules and supply agreement?
- What may the brand pass on to its own customers, retailers or regulators? What stays confidential?
- What happens to the documents if the business relationship ends?
Change notification
- Which changes need advance notice: formula, raw-material supplier, method, site, equipment, packaging component, specification?
- How much notice, and does the brand need to approve before the change takes effect?
- How will the specification revision history be shared?
Exceptions and escalation
- Who is the named contact on each side for OOS results, deviations and complaints?
- Within what period will the brand be told about an OOS result on its batch?
- What is the escalation path if the first-level contacts cannot resolve an issue?
Late or incomplete documents
- If the COA is late, does the shipment wait, ship on hold, or ship with the brand's written agreement?
- If an item is missing or reported only as "pass", what is the correction route?
- Can goods go into the brand's stock or onward sale before documents are complete? Who decides?
FAQs
What can a batch COA show? It shows that a specific batch was tested against stated items, using stated methods, with stated results, at a stated specification revision. It covers only that batch and only those items.
What can it not establish by itself? It cannot establish shelf life, preservative efficacy, packaging compatibility, consumer safety, legal compliance or market approval on its own. Some measurements on a COA may contribute to a wider assessment, but those conclusions need their own evidence and, where relevant, regulatory steps in the target market.
Should every specification item appear on the batch COA? No. Formula-level properties such as challenge test results and stability are usually backed by separate reports. The specification should name the evidence source for each item so that nothing is left uncovered.
Is a raw-material COA enough for a finished product? No. It describes an incoming ingredient lot from that ingredient's supplier. It supports traceability but says nothing about how the finished batch performs.
How should third-party lab reports be treated? Check the sample identity, batch or formula code, method, date and scope. A report on a different batch, older formula or different pack does not transfer automatically; a documented impact assessment may support an appropriate bridge or show that new work is needed.
What if a result is missing or shows only "pass"? Ask for the numeric value, units and method wherever the agreed COA scope calls for a numeric result. A conform/not-conform statement can serve an agreed qualitative item, such as appearance against a reference standard. Decide any exception against the approved specification and quality agreement.
Why request the actual value for a numerical item? "Pass" gives no information about where a result sits within the range. The actual value helps compare batches tested by equivalent methods and investigate a complaint later. That reporting expectation should be agreed before production.
What happens when the specification changes? Record the approval required under the agreement, the effective rule and the first affected batch in writing. COAs should cite the revision applicable to their own batch. Assess whether any separate studies still apply. Case B shows what happens when a COA cites a revision that may be wrong for that batch.
Can the brand set limits on its own? The brand can propose limits. Final limits should be agreed from development data, method capability and verified target-market requirements, with the relevant quality and regulatory specialists involved.
Closing Checklist: What to Send When Requesting a Project-Specific Specification and Quote
- Product brief: product type, format, intended use, target user, claims under consideration
- Target markets and sales channels for regulatory review
- Formula status: the brand's own formula, the manufacturer's base formula, or a new development
- Pack details: primary container, closure, dispenser, material, fill size
- Specification items the brand expects, with proposed ranges marked as proposals
- Request to name the method and evidence source (batch COA, separate report or raw-material control) for each item
- Separate studies wanted: challenge test, stability, compatibility, third-party tests, and who owns each
- COA expectations: numeric results, revision citation, signatory, format, recipient
- Questions from the pre-PO list above that need answers before ordering
- Forecast volumes and batch-size range, so the manufacturer can quote. Leave MOQ and lead time for the manufacturer to state
- Required document set for each delivery, and the brand's review contact
Summary
The approved finished-product specification sets parameters, methods and limits for a defined formula version, and records which items are batch-release tests. A batch cosmetic certificate of analysis reports actual results for those agreed items, against the specification revision in effect for that batch. Laboratory reports cover specific studies, and GMP certificates describe a facility. Raw-material COAs describe ingredients, not your finished product. A COA is one quality record and does not by itself show safety, compliance or market approval.
Next step
Before the first order, ask your manufacturer for a project-specific finished-product specification and quotation, including which items the COA will report. For OEM production of a locked formula the brand already owns, review our OEM manufacturing service and build the specification around that exact formula version. For ODM development, agree how the specification will be drafted and approved once the formula is final. Agree which studies are in scope, when and in what format the COA will be delivered, then request a project quotation with the target market, packaging, desired COA fields and an estimated order-quantity range.
Ready to Start Your Project?
Tell us your product requirements. Within 24 hours, our technical team will reply with a quotation or the questions needed to prepare one.