How to Set Perfume QC Specifications | Appearance, Fill Volume, Scent Match

Perfume QC limits should come from the actual product and production process. Use an approved master sample, real filling and test data, packaging checks, stability results, and the rules that apply in the sales market. For finished perfume, the main checks usually include appearance, color, fill volume, scent match, density where useful, bottle and pump condition, leakage, artwork, and batch traceability. ISO 22716 provides GMP guidance for cosmetic production, control, storage, and shipment.^[1]

Each item on the QC sheet should say how it is tested, what it is compared with, what passes, how much is sampled, and what to do if it fails. Avoid loose wording such as “normal color,” “correct smell,” or “acceptable fill.” Different inspectors may read those words differently.

Master Sample

Use an approved commercial bottle as the main QC reference. It should match the formula, fragrance concentration, bottle, pump, cap, and fill size used for production. An early lab sample is not enough. XUELEI's perfume manufacturing process also links formula approval, bulk production, filling, packaging, and finished-product QC.

Keep one working bottle for daily checks and at least one sealed archive bottle. The working bottle will slowly change each time it is opened. Air, light, heat, and evaporation can affect both color and smell.

• Working reference: opened for routine comparison.
• Archive reference: kept sealed so QC can check whether the working bottle has changed.

If the working bottle becomes darker, weaker, or starts smelling different from the sealed archive sample, replace it. Do not change the QC limit to match an old reference bottle.

Review the master sample after a formula change, fragrance supplier change, concentration change, carrier change, major raw-material change, or a change to the bottle, pump, or another part that touches the perfume.

Set the QC Limits

Do not copy a number from another perfume and use it as the new product limit. Start with data from approved samples and actual production.

1. Measure approved development or production batches.
2. Record the normal range and normal variation.
3. Make sure those batches also pass scent, packaging, and stability checks.
4. Look for the point where normal variation starts to become process drift.
5. Set the production target, warning point, and final acceptance criteria where needed.

These numbers do not have to be the same. A filling machine, for example, may be set slightly above the amount printed on the bottle so normal machine variation does not create too many low fills. That machine setting is a production target, not automatically the legal pass/fail rule.

The same idea works for density and color. If approved batches repeatedly stay inside a narrow range, that history gives a useful starting point for the QC limit. FDA cosmetic GMP guidance also covers manufacturing controls, equipment, records, and inspection during production.^[2]

Appearance

For a clear perfume, record clarity, particles, haze, sediment, and separation separately. A single “appearance: pass” result hides too much information.

Clarity

Clear and homogeneous, matching the approved visual reference.

1. Use the same neutral inspection light.
2. Check the bottle against a white background.
3. Check it again against a dark background.
4. Hold it upright and turn it slowly.
5. Look at the shoulder and the bottom.
6. Wait for filling bubbles to disappear before judging haze.

A dark background makes pale fibers easier to spot. A white background helps with dark particles and some color differences. Do not shake the bottle just before inspection because bubbles can look like particles or haze.

For products with tight visual standards, keep the inspection distance, lighting, viewing angle, and inspection time the same. An internal setup might use one inspection station and a viewing distance of around 30-50 cm. That is only an example, not a general perfume standard.

Defect location matters as much as defect size. A 0.5 mm mark under the bottle and the same 0.5 mm mark across the front logo may need different defect grades.

Particles

No readily visible foreign particles under the approved inspection method.

Look for dust, fibers, glass pieces, plastic debris, pump debris, insoluble material, bulk sediment, and contamination picked up during filling.

If the same type of particle appears in several bottles, look for the common source. Removing one particle from one bottle does not tell QC whether the rest of the batch is clean.

Haze and Sediment

If a perfume was approved as clear and later shows persistent haze, hold the batch and check the water level, fragrance solubility, temperature, raw materials, filtration, formula ratio, and possible contamination.

If a slightly hazy look is normal for a particular formula, keep a physical sample showing the maximum acceptable haze. The words “slight haze” are not enough on their own.

For sediment, check precipitation, insoluble ingredients, contamination, oxidation, filtration, and ingredient compatibility. If several finished bottles show the same problem, compare them with a bulk sample. That helps show whether the issue started before or during filling.

Do not pass a perfume just because shaking makes the sediment disappear for a short time.

Separation

A perfume approved as one clear phase should not show an oil layer, droplets, two liquid layers, floating material, or repeated phase separation.

If this happens, hold the batch and check material quantities, mixing, transfer conditions, tank conditions, and production records. Shaking the bottle before inspection does not solve a product that separates again later.

Color

For visual color control, keep three physical references: a center sample, a light-limit sample, and a dark-limit sample.

• Match: close to the center sample.
• Acceptable boundary: still between the approved light and dark limits.
• Reject: outside the approved range.

The same method works well for scratches, coating defects, print position, cap alignment, and haze. A physical boundary sample is easier to use than terms such as “slight” or “minor.”

Instrumental Color

If visual checking is not precise enough, measure L*, a*, and b* values using the same method each time.

Do not take a ΔE limit from another perfume. Compare approved production samples with samples that the brand has already rejected.

These are sample numbers only. They show how approved and rejected product data can be used together. They do not mean that ΔE 1.2 is a general pass limit for perfume.

Keep the sample container, instrument setup, optical path, and temperature the same. Also look at the perfume through the final bottle. A color difference that is easy to see in clear glass may be much less obvious in smoked, colored, coated, or opaque packaging.

Color Change

Do not stop color checking at Day 0. A perfume may match the approved color when it is filled and become darker later.

XUELEI's information on fragrance formulation, testing, and production also includes stability work before commercial production.

If products close to the dark limit repeatedly move outside the approved range during storage, check the release limit, formula, light exposure, packaging, and other possible causes of color change.

Fill Volume

Use a measured net-content method instead of judging the liquid level by eye. Decorative glass bottles can have slightly different internal shapes and wall thicknesses, so equal volumes do not always sit at the same height.

Measure by Weight

Net perfume mass = filled package weight − empty package tare

Volume = net perfume mass ÷ perfume density

• Filled bottle: 242.60 g
• Empty package tare: 157.10 g
• Net perfume mass: 85.50 g
• Perfume density: 0.855 g/mL

85.50 ÷ 0.855 = 100.0 mL

Use the density of the actual perfume under a set test condition. Do not assume every alcohol-based fragrance has the same density. XUELEI's perfume concentration guide shows how the fragrance and carrier levels change between different perfume types.

Control Tare Weight

Check how much the empty bottles vary before using one average tare across the whole packaging lot.

• QC tare: 157.10 g
• Actual empty bottle: 158.00 g
• Tare error: 0.90 g

At 0.855 g/mL density:

0.90 ÷ 0.855 ≈ 1.05 mL

In this example, a 0.90 g tare error changes the calculated fill by about 1.05 mL.

Sample empty bottles from each packaging lot, calculate the average tare and the spread, and update the tare when the bottle lot changes. For tighter control, weigh each bottle before and after filling.

Nominal Quantity, Target, and Acceptance

The filling target may sit above the nominal quantity because filling equipment naturally has some variation. One bottle slightly below the label amount does not automatically mean the whole lot fails.

In the United States, NIST Handbook 133 uses both an Average Requirement and an Individual Package Requirement for packaged goods.^[3]

For prepackages covered by the EU harmonized system, Directive 76/211/EEC separates nominal quantity, actual contents, negative error, and tolerable negative error. Volume checks under the Directive are made at, or corrected to, 20°C.^[4]

Do not use a blanket rule such as “always overfill by 2%.” Use data from the actual filler, bottle, perfume, measuring method, and sales market.

Pilot Fill Data

Run enough bottles during validation to see how the filler really behaves before setting the normal target.

These numbers are only an example. Real limits should come from the actual filling machine, bottle, perfume, and measurement method.

Process Variation

Line B has the higher average, but it also has much wider variation. That can lead to both excessive overfill and low-fill bottles.

Track the average, minimum, maximum, spread or standard deviation, individual nozzle results, and the trend over time.

A sequence of 50.7 → 50.6 → 50.4 → 50.2 mL shows a clear downward trend even if every result is still inside the current internal limit. This is where a warning limit helps.

Measurement Error

Check scale resolution, calibration, repeatability, tare method, density method, test temperature, and operator instructions.

If the same bottle gives readings that move almost as much as the entire QC tolerance, the measuring method needs to be fixed before those numbers are used for release decisions.

Overfill Cost

That 1,000 L is the same amount of liquid used for 10,000 additional 100 mL bottles.

If unnecessary overfill on one million bottles drops from 1.0 mL to 0.4 mL, the reduction is 600 L of finished perfume.

In-Process Fill Checks

Check fill quantity at startup, after adjustments, after long stops, after bulk replenishment, after nozzle changes, at set intervals during the run, and near the end of filling.

For multi-nozzle machines, record each nozzle separately. One low-running nozzle can disappear inside a good-looking line average.

Scent Match

Set a fixed scent-check method with one master reference, one blotter type, the same spray method, fixed evaluation times, selected scent attributes, and a clear pass/fail rule.

Test Setup

1. Code the reference and production samples.
2. Use the same blotter type.
3. Apply the same number of full sprays.
4. Keep the spray distance and method the same.
5. Hide the sample identity where practical.
6. Check both samples at the same times.

Use the same testing surface for both samples. Skin testing can be useful during development, but skin temperature, creams, soap, and personal skin condition add extra variation. A standard blotter is easier to control for routine batch QC.

Scent Attributes

In this example, Batch A stays close to the reference. Batch B changes the balance: citrus drops from 7/10 to 4/10 while sweetness rises from 4/10 to 6/10.

Choose attributes that suit the perfume. A citrus scent may focus on brightness, freshness, green notes, and sharpness. A gourmand scent may focus on vanilla, sweetness, creaminess, caramel, and woody notes.

ISO 13299 gives guidance for building sensory profiles, including products assessed by smell such as cosmetics.^[5]

Scent Development

The opening matches in this example, but the difference is clear after 60 minutes. This is why checking only the first spray is not enough.

Pass, Borderline, Fail

• Pass: no meaningful repeatable difference from the master sample and no unacceptable off-note.
• Borderline: a small repeatable difference is found and needs the agreed review process.
• Fail: a repeatable clear difference, missing key note, wrong balance, or unacceptable off-note is found.

Do not use a rule such as “every score must stay within ±1 point” unless testing on that perfume has shown that the number really matches the brand's pass/fail boundary.

Strength vs. Character

Do not record only “odor fail.” Write what actually changed: too weak, too strong, balance changed, key note missing, unexpected note, off-note, unusual development, or different dry-down.

Off-Notes

Useful descriptions include sour, metallic, burnt, stale, oxidized, unusually alcoholic, unusually sweet, flat, weak opening, and missing freshness.

A perfume can still smell like the correct product and have an unacceptable off-note.

If the product needs a set resting period before final scent inspection, include that time in the approved production and QC procedure. Do not add a waiting period only after a batch fails.

Evaluator Conditions

• Do not wear strong personal fragrance during testing.
• Keep strong background odors out of the room.
• Take breaks instead of smelling many strong samples continuously.
• Postpone formal testing when an evaluator has obvious nasal congestion.

Use more than one trained evaluator for important products. After repeated exposure to strong perfume, the nose can become less sensitive to differences.

Triangle Test

In a triangle test, the assessor receives three coded samples. Two are the same and one is different. The assessor has to identify the different sample.

ISO 4120:2021 gives the formal triangle-test method and notes that it has limited use for products with strong carryover or lingering sensory effects.^[6]

This matters for perfume. Control sample order, coding, exposure, breaks between samples, room odor, the number of assessors, and the final decision rule. One correct answer from one person does not prove that two batches are different.

Batch Homogeneity

During validation or when uneven mixing is suspected, compare samples from the beginning, middle, and end of the filling run.

These numbers are only examples. The second set does not prove that mixing failed, but it gives a good reason to check the bulk tank, mixing, transfer, settling, sampling, and long filling stops.

Instrument Testing

GC or GC-MS can help check for missing or unexpected compounds, raw-material changes, oxidation, the wrong fragrance material, or a formula difference. It is most useful as part of an investigation, not as an automatic replacement for scent testing.

Density and pH

Density can help flag a wrong alcohol/water ratio, incorrect fragrance dosage, wrong bulk material, mixing errors, or a problem in the fill-volume calculation.

For example, if approved batches repeatedly measure between 0.852 and 0.857 g/mL using the same method and temperature, that history may support an internal density range for that formula. This is an example, not a general perfume standard.

If density fails, check the test temperature, sample identity, formula quantities, carrier ratio, material identity, and test method before making any change to the batch.

Do not add pH simply because the product is a cosmetic. Many fine fragrances are mainly alcohol-based, so a normal water-based pH test may not give useful QC information. Use pH only when it makes sense for the formula and the test method has been defined.

Bottle, Pump, and Closure

Check the finished package together with the perfume. XUELEI's fragrance OEM/ODM manufacturing workflow covers fragrance, packaging, filling, and finished-product production as linked parts of the same product.

Bottle

• Cracks and chips
• Deep scratches
• Foreign material
• Severe glass distortion
• Unstable base
• Coating or paint defects
• Wrong glass color
• Decoration errors
• Incorrect logo position

Judge visible defects by both size and location. A scratch across the front logo may be major, while the same-size mark under the bottle may be minor. Keep approved defect samples or clear reference images.

Pump and Spray

Check priming, spray shape, dose output, actuator return, leakage, alignment, and dripping.

1. Weigh the bottle.
2. Apply 10 full sprays.
3. Weigh it again.
4. Calculate the mass loss.
5. Divide by 10.

If 10 sprays use 1.00 g:

Average output = 0.10 g per spray

This number is only an example.

Do not look only at the average. A pump that swings sharply from one spray to the next gives a much less consistent user experience.

Crimp and Leakage

For crimped pumps, check crimp dimensions where needed, pump alignment, collar condition, visible deformation, and leakage. Use limits from the actual bottle-and-pump combination rather than one setting for every package.

Leakage testing can include side storage, upside-down storage, elevated temperature, temperature cycling, transport simulation, and suitable vacuum or pressure tests.

During normal QC, look for liquid around the neck, liquid inside the cap, stains around the pump, measurable product loss, or a strong perfume smell around a sealed bottle or carton.

Cap and Artwork

Check cap identity, fit, alignment, retention, decoration, scratches, looseness, and magnetic orientation where relevant.

Check the approved artwork for:

• product name;
• nominal quantity;
• barcode;
• ingredient information;
• batch identification;
• required warnings;
• market-specific information;
• print quality;
• label position.

Use controlled artwork files with the correct revision, approval date, market, and SKU. Do not use a screenshot from email or a messaging app as the final QC reference.

For cosmetics sold in the EU, Regulation (EC) No 1223/2009 contains requirements covering safety, GMP, responsible-person duties, product information, ingredients, and labeling.^[7]

Defect Levels and Sampling

Look at frequency as well as severity. One small mark may be minor. The same mark on many bottles may point to tooling damage, conveyor contact, packaging damage, or supplier drift.

Sampling

Set sample size based on lot size, risk, production history, test type, legal requirements, customer requirements, and whether the test destroys the sample.

Use one written sampling plan for comparable lots. Do not inspect five units from one lot and twenty from a similar lot unless there is a clear reason.

AQL

AQL-type sampling can be used for suitable visual and packaging defects such as scratches, print problems, decoration defects, cap defects, and packaging damage.

Do not use one AQL plan in place of net-content checks, formula verification, scent approval, or critical safety controls.

Perfume QC Specification Example

Add the sample size, test frequency, reason for each limit, actual result, and action after failure to the working QC sheet.

50 mL QC Example

The pilot numbers above are examples only. Final limits should come from the actual 50 mL SKU, formula, filler, package, and market requirements.

Failed Results

Hold → Confirm → Investigate → Decide

Hold

Block the affected batch and record the batch number, quantity, location, failed test, and date.

Confirm

Check the sample and test method before repeating anything. Review the tare, scale, calibration, temperature, scent reference, blotter, and sample identity.

If the first density result fails and the next two pass, do not release the batch simply because two out of three results passed. The first failure still needs an explanation.

Investigate

If the batch may not be uniform, compare samples from the beginning, middle, and end of the run.

Decide

Use a recorded decision: sorting, controlled rework, refill, component replacement, partial rejection, or full-batch rejection.

After rework, repeat every test that may have been affected. A fill correction, for example, can also affect the closure, leakage, cleanliness, coding, and packaging.

Retain Samples and Stability

Keep sealed finished-product samples from each released batch. They give QC an untouched reference if a complaint appears later.

Record the batch, production date, retained quantity, storage place, storage condition, and retention period.

This table is only an example. It shows how a product can pass on Day 0 and develop a problem later.

Track scent, color, clarity, precipitation, separation, evaporation, bottle coating, labels, pumps, caps, and leakage. Do not treat a short accelerated test on its own as proof of a fixed shelf life.

Supplier and Change Control

Record the identity and revision of incoming fragrance concentrate, bottles, pumps, caps, and other materials that touch the perfume.

For fragrance concentrate, record the supplier, fragrance code, batch, revision, required documents, certificate of analysis where used, and applicable IFRA documents.

For bottles, pumps, and caps, record the supplier, component code, drawing, dimensions, material, decoration, and agreed defect limits.

An IFRA Certificate of Conformity is not the same as full regulatory approval. IFRA states that compliance with its Standards does not replace a finished-product safety assessment or the need to meet national and local rules.^[8]

XUELEI's ODM fragrance manufacturing framework also links formula, packaging, stability, production, and QC because one supplier or component change can affect several finished-product checks.

Changes That Need Review

Release Checklist

• Correct formula and fragrance code
• Appearance within approved limits
• No unacceptable particles
• Color within approved boundary
• Scent match passed
• Density passed where required
• Fill results passed
• In-process fill records checked
• Bottle and pump passed
• Crimp and leakage passed
• Cap passed
• Correct artwork revision
• Correct batch code and barcode
• All deviations closed or formally approved
• Retention samples stored
• Authorized batch release completed

Even a passing result should be checked when it moves sharply away from normal production history.