Resource 02 · Research & Development

Formulation Development Framework

How do we transform a product brief into a stable, safe, reproducible and scientifically defensible formulation?

Educational framework
Edition
1.0
Last technical review
18 November 2026
Full methodology
12 min read

Developed and curated by Shamim Quresh · Editorial methodology

Level 1 · Understand · about two minutes

The framework at a glance

A formulation is a designed physicochemical system. Its ingredients, process, packaging and storage environment together decide how it behaves. Attractive ingredients do not make a good product, and a beautiful laboratory batch is not evidence that a factory can make it, that it will survive the final pack, or that it does what the label says.

Controlling question
Can it reliably meet the product's specified performance and safety requirements?
Key decision record
Approved development report and master formulation
Role in the library
Establishes the technical product
Applies to
Cosmetic and personal-care formulation. The scientific reasoning applies widely; the regulatory examples depend on product classification and market. Sunscreens and products with medical claims follow additional rules in many markets.

The 12 steps

Step 01 of 12

Target product profile

Turn the brief into measurable targets.

Essential deliverables

  • Target product profile
  • Requirements-to-test matrix
  • Ingredient-function matrix
  • Experimental batch records
  • Stability and compatibility protocol and results
  • Claim substantiation
  • Laboratory-to-factory transfer report
  • Approved master formula and change rules

Common mistakes

  • Adding more emulsifier to every instability.
  • Changing several variables at once and calling the result an experiment.
  • Treating centrifuge or heat-stress screens as shelf-life proof.
  • Testing in glass and launching in a different pack.
  • Scaling by multiplying every quantity by 100.
  • Evaluating 'no white cast' on a single light skin tone.

After reading, you should be able to

  • Turn adjectives in a brief into measurable requirements with tests.
  • Choose a formulation architecture and name its dominant stability mechanisms.
  • Plan screening, optimisation and confirmation experiments rather than changing several variables at once.
  • Design a stability and compatibility programme in the final pack.
  • Identify what must transfer from laboratory to factory.

How it connects to the other frameworks

Level 2 · Study · the complete methodology

Work the method

Purpose and scientific philosophy

A formulation is not a list of attractive ingredients. It is a designed physicochemical system whose components, process, packaging and storage environment collectively determine its behaviour. “Natural” does not inherently mean safe; “clinically studied ingredient” does not establish performance of the finished formula; and apparent laboratory stability does not replace a defined, justified stability programme.

Twelve-step development workflow

1. Translate the brief into a target product profile

Record intended user, application site, use frequency, sensory expectations, format, claims, target markets, consumer price, packaging, shelf-life intention, microbiological risk, environmental conditions, manufacturing batch scale and unacceptable outcomes. Turn adjectives into measurable targets: “lightweight” becomes a defined sensory panel rubric and, where relevant, a viscosity or spreadability range. “Non-whitening on deeper skin tones” requires a representative shade-diverse evaluation plan, not a single swatch on a pale arm.

2. Create a requirements-to-test traceability matrix

RequirementScientific mechanism or riskTest/evidenceAcceptance decision
Uniform appearanceDispersion/emulsion stabilityVisual, microscopy, centrifuge screening and real-time observation as applicableDefined absence of unacceptable separation/agglomerates
Intended viscosityFlow and dispensing behaviourValidated viscometer conditions, shear profile if neededProduct-specific range and dispensing suitability
Consumer sensory experienceSlip, drag, absorption, residueBlinded, structured panel across relevant skin typesPredefined scores, variability and confidence limits
Microbial qualityWater activity, contamination and preservationMicrobial limits and preservation efficacy where appropriateApplicable method/standard criteria
Packaging compatibilitySorption, leaching, corrosion, closure integrityFilled-pack stress/storage observations, targeted analytical testingNo unacceptable change to product or pack
Chemical stabilityOxidation, hydrolysis, photolysis, pH driftStability-indicating analyses where feasiblePre-established critical quality limits
Claim supportRelevant performance endpointAppropriately designed finished-product or defensible bridging evidenceClaim-specific substantiation standard

Do not invent universal numerical acceptance criteria. They depend on product, risk, method, standard and intended use.

3. Map the ingredient system by function, not marketing appeal

Build a composition table showing INCI or appropriate ingredient identity, grade, supplier, function, proposed concentration, known compatibility constraints, impurity concerns, solubility, processing needs, permitted-use status in each destination market and evidence status. Distinguish functional materials (solvents, surfactants, emollients, emulsifiers, rheology modifiers, preservatives, chelators, antioxidants) from claimed actives. Multiple functions can coexist, and interactions matter more than isolated ingredient reputations.

4. Conduct a hazard, compatibility and regulatory pre-screen

Screen ingredient restrictions, impurities, allergens, foreseeable use and exposure, compatibility with pH, ionic strength, oxygen, light, temperature and packaging. For sunscreen, recognise that UV filters, SPF/UVA testing, classification and approvals can be legally distinct from ordinary cosmetics, depending on market; no SPF or broad-spectrum claim follows from a plausible ingredient list alone.

5. Select a formulation architecture

Choose the physical system: solution, gel, oil, emulsion, suspension, anhydrous balm, aerosol or another defined form. Make the architecture choice explicit: why this system, which stability mechanisms dominate, and which trade-offs are accepted? Examples: polymer-thickened gels can be sensitive to electrolytes; emulsions depend on interfacial stabilisation and process history; pigment-containing systems require suspension and dispersion control; oxygen-sensitive ingredients may demand package-level protection.

6. Plan experiments before preparing batches

Begin with a screening design to identify influential variables, then a targeted optimisation design to understand interactions. Variables might include emulsifier level, oil-phase fraction, mixing intensity, temperature, cooling profile and pH adjustment. Responses might include viscosity, droplet distribution, separation propensity, skin feel and chemical retention. Randomise run order when practical, control material lots and document nuisance variables. A fractional factorial design can reduce trial count, but its confounded effects must be acknowledged. Do not optimise exclusively for the nicest-feeling fresh batch.

7. Manufacture controlled laboratory prototypes

Use a standard batch record, calibrated equipment, controlled ingredient identification and complete process parameters. Record order of addition, mixing geometry, rpm or tip speed where relevant, duration, temperature, batch size, hold times, pH adjustments, observations, deviations and sample IDs. Photograph or instrumentally capture relevant observations. Each prototype gets an immutable ID; never silently overwrite a formula revision.

8. Screen quickly, then investigate deeply

Use low-cost, high-information screens early: microscopy, appearance, pH drift, viscosity response, centrifugation where scientifically appropriate, thermal cycling and preliminary package observations. These are risk screens, not standalone shelf-life proof. Investigate surprising behaviour mechanistically rather than adding ingredients at random.

9. Test stability, preservation and package interactions

Define a written protocol for representative packaging, storage conditions, sample frequency, analytical methods, acceptance criteria and responsibilities. Include real-time storage and justified accelerated/stress conditions, recognising that acceleration does not always predict real-time behaviour. Assess microbial risk and preservation efficacy as applicable. Evaluate fill/closure integrity, dispensing, weight loss, compatibility and possible extractables/leachables according to risk. A compatible bulk sample can fail in the final pack.

10. Verify intended performance and substantiate claims

Choose methods that actually measure the claim. A hydration claim may require an appropriate instrumental or clinical endpoint; a “brightening” claim needs a defined endpoint and population. Specify sample size rationale, inclusion criteria, bias controls, statistical plan and adverse-event handling when relevant. Match the strength and wording of the claim to the evidence. For skin-tone inclusivity, include relevant melanated skin populations and realistic use conditions rather than treating inclusion as imagery alone.

11. Transfer to manufacturer and confirm scale representativeness

Identify scale-dependent variables: mixing energy per volume, tip speed, heat transfer, addition rate, vessel geometry, deaeration, filtration, transfer-pipe effects and filling shear. Conduct a witnessed trial, compare laboratory and factory samples against the same critical quality attributes and investigate differences. Do not substitute “multiply every ingredient by 100” for process transfer. Define sampling locations, hold times, rework rules and cleaning considerations.

12. Close the development file and control change

Final outputs: approved master formula, raw-material specifications, manufacturing instructions, process ranges, analytical/sensory methods, product specification, stability data and ongoing commitments, preservation/microbiology evidence, packaging specification, claim substantiation, safety-assessment input dossier, approved label content and change-control rules. Record what remains unverified; do not describe an open test as passed.

Illustrative study — tinted product for deeper skin tones

Question: Can a tinted, high-solids topical product maintain shade uniformity, dispensing performance and comfortable feel across realistic storage conditions? (This example does not claim sunscreen efficacy or safety.)

Hypotheses: Pigment wetting and dispersion energy affect agglomeration; dispersant selection and oil-phase structure affect sedimentation; package geometry and yield stress affect ease of dispensing; pigment appearance varies with substrate tone and lighting.

Experiment: Compare a small matrix of dispersion processes and structuring approaches, with controlled pigment lot and total pigment loading; measure dispersion through microscopy or validated particle analysis, flow under relevant shear conditions, shade consistency on diverse skin-tone standards, and pack dispensing after defined storage intervals. Blind sensory evaluation where feasible. Then repeat promising conditions with a second ingredient lot and at manufacturing-relevant scale.

Decision: Select the lowest-complexity option that meets predefined acceptance criteria. A formula that looks elegant in the beaker but plugs a closure is not successful. Any SPF/UVA, water-resistance or medical claim requires its own market-specific test and regulatory pathway.

Creative and efficient methods

  • Mechanism map: Draw every failure as a chain — trigger → physical/chemical mechanism → observable symptom → confirming test → control. This prevents “more emulsifier” from becoming a reflexive response to every instability.
  • One-problem/three-hypothesis rule: Before changing a formula, propose at least three plausible causes and one test that distinguishes them.
  • Material substitution stress test: Challenge the design with a second qualified supplier or lot early. Fragility to a tiny raw-material variation is a scale-up warning.
  • Split-lot packaging trial: Fill the same prototype into competing packages to isolate package interaction from formula change.
  • Maximum information per gram: Use microbatches for early physicochemical screening, reserving larger batches for filling/scale-dependent evaluation where microbatches cannot represent the process.
  • Negative result library: Catalogue failed formulas with cause hypotheses, conditions and evidence. Failed experiments are reusable assets when they remain searchable.
  • Decision-weighted DoE: Rank experiments not only by scientific interest but by how much each result could change a costly project decision.

Laboratory-to-manufacturer transfer checklist

Before transfer, verify exact ingredient identities and grades, acceptable substitutions, tolerances, batch instructions, in-process controls, product specifications, packaging requirements, cleaning considerations, approved sample retention, deviations/CAPA process, ownership of intellectual property and records, quality agreement, change notification and scale-up responsibilities. Require a formal transfer report with results and unresolved issues.

JUA — a fictional technical case study

Brief. A tinted daily facial sunscreen for adults in strong-sun climates, designed to sit well on deeper skin tones, packed in a brushed aluminium tube.

Where the framework bites first.

AreaThe question to answerHow it would be investigated
ClassificationWhich market, and is a sunscreen a cosmetic or a drug there?Requirements register per market before formulation locks. In the United States sunscreens are regulated as drugs; in the EU they are cosmetics with UV filters restricted to an approved list
Visible cast on melanated skinDo the chosen UV filters and pigments leave a grey or white cast?Shade-diverse evaluation plan on a defined range of skin tones and lighting conditions, not a single swatch
Pigment distributionDo iron-oxide pigments stay evenly dispersed, and does the shade drift with storage?Microscopy or particle analysis; colour measurement on fresh and stored samples; dispersion energy as an experimental variable
RheologyIs the product thick enough to hold pigment in suspension but thin enough to dispense and spread?Flow curve across relevant shear rates; yield-stress estimate; dispensing test through the actual tube orifice
Aluminium tube compatibilityWill the product attack the internal lacquer lining, or the metal through a lining defect?Filled-tube storage at justified conditions; cut-open inspection of the lining; check for discolouration, gas, corrosion, weight loss and changes in pH or viscosity
DispensingDoes the tube behave well to the last gram?Aluminium tubes do not spring back, which limits air ingress but changes how the last portion dispenses; test with consumer handling
StabilityDo appearance, shade, viscosity, pH and filter content stay within limits?Written protocol with real-time and justified stress conditions in the final pack
Scale-upDoes factory dispersion and filling reproduce the laboratory product?Witnessed transfer trial comparing the same critical quality attributes; fill temperature, deaeration and shear during transfer recorded
Testing strategyWhich claims, which methods, which laboratories?SPF and UVA testing to the methods required in each market (for example ISO 24444 for in vivo SPF and ISO 24443 for in vitro UVA, where accepted); claims only after results exist

Decision rule. Choose the simplest formulation and pack combination that meets every predefined acceptance criterion in the final pack. Any protection claim waits for the market-specific evidence.

Working documents

Target product profile

FieldExample of a measurable entry
Intended user and useAdults; face; once daily and on reapplication
Format and packTinted cream; 50 mL aluminium tube with lined interior
Sensory targetsDefined panel rubric for spreadability, residue and finish
AppearanceShade range defined against a reference set of skin-tone standards
Claims (hypotheses)Listed with the evidence each would require — not yet approved
MarketsListed with the classification in each
Shelf-life intentionTarget, to be confirmed by the stability programme
Unacceptable outcomesVisible cast on deeper skin, separation, tube corrosion, pilling under make-up

Ingredient-function matrix

FunctionWhy it is thereQuestions to record
Solvent / continuous phaseCarries and distributes the systemGrade, water quality, microbial risk
EmollientsSkin feel, spreading, pigment wettingPolarity, compatibility with filters and pigments
EmulsifiersStabilise the oil–water interfaceSystem type, sensitivity to electrolytes and process history
Rheology modifiersStructure, suspension, dispensingShear behaviour, salt and pH sensitivity
Pigments / colourantsShadeDispersion route, coating, permitted status per market
PreservativesMicrobial protectionSpectrum, pH window, packaging and partition effects
Chelators / antioxidantsProtect against metal-catalysed and oxidative changeInteraction with metal packaging and pigments
Claimed activesThe reason for a claimEvidence for the finished formula, not just the ingredient

Each row also carries INCI name, supplier, grade, proposed concentration, permitted-use status in each destination market and evidence status.

Experimental batch record

Record for every prototype: prototype ID and formula revision; date and operator; raw-material lots; equipment and mixer geometry; batch size; order of addition; temperatures, mixing speeds and durations; hold times; pH adjustments; in-process observations; deviations; samples taken and their IDs; photographs or instrument files. Never overwrite a revision.

Stability testing plan

ElementWhat to define before the first sample goes in
SamplesFormula revision, batch, pack (the final pack wherever possible)
ConditionsReal-time and justified stress conditions relevant to the intended climates
Time pointsSchedule and the decision each time point informs
Attributes and methodsAppearance, colour, odour, pH, viscosity, separation, pack interaction, active or filter content where measurable
Acceptance criteriaProduct-specific limits set before testing
ResponsibilitiesWho pulls, tests, reviews and decides
InterpretationHow out-of-limit results are investigated; what stress results can and cannot predict

Compatibility assessment

Fill the same prototype into each candidate pack (a split-lot trial). Compare after storage: product appearance and properties, pack appearance, closure integrity, weight loss, dispensing and, where the risk justifies it, targeted analysis for migration.

Formulation revision log

RevisionDateChangeReasonEvidenceApproved by
FRM-001-R01—Initial prototype—Batch record—
FRM-001-R02—One controlled changeHypothesis testedResult linked—

Failure investigation guide

Each scenario has several plausible causes. Propose at least three, choose a test that tells them apart, and only then change the formula.

SymptomPlausible causes to discriminateDiscriminating evidenceTypical corrective direction
Emulsion separationInterfacial film too weak; droplet size too large; electrolyte or pH shift; process temperature or shear historyMicroscopy over time; droplet size; pH and conductivity; comparison of batches made with different process historiesAdjust emulsifier system or process, not reflexively "more emulsifier"
Viscosity driftPolymer hydration incomplete; neutralisation timing; temperature history; raw-material lot; measurement set-upViscosity over time at controlled temperature; retained raw materials; method checkFix the process step or specification that varies
Pigment sedimentationInsufficient yield stress; poor wetting or dispersion; flocculationSettling observation; microscopy; flow curveImprove dispersion, dispersant or structure; confirm dispensing still works
pH instabilityHydrolysis; buffer capacity too low; interaction with pack or pigmentpH trend in glass versus final pack; buffer capacity checkReformulate buffer system or isolate the reacting component
Oxidation and discolourationOxygen, light or metal-catalysed oxidation; ingredient impurityHeadspace and pack comparison; light-protected controls; chelator comparisonAntioxidant or chelator strategy; pack protection; supplier specification
Packaging incompatibilityLining defect; sorption into plastic; stress cracking; closure leakageSplit-lot pack trial; cut-open inspection; weight lossChange pack component or formula; requalify
Preservation failurePreservative outside its pH window; partition into oil phase; inactivation by other ingredients or packPreservation efficacy testing; pH check; formula reviewRedesign preservation system; retest; never release on assumption

Level 3 · Apply · educational tools

Practise the method

These tools are teaching simulations. They use illustrative data, show their assumptions and are not validated scientific, regulatory or financial instruments.

Conceptual demonstration

Formulation laboratory

Choose a physical system to see its structure, what each class of ingredient does, what keeps it stable, which process variables matter, how it fails, how it is tested and what the pack must cope with.

Circles: droplets; outline: emulsifier film at the interface. Schematic, not to scale.

Qualitative bench: how fast would droplets cream or sediment?

For an isolated sphere in a simple liquid, Stokes' law gives speed = 2 r² Δρ g ⁄ 9η. Change one factor at a time and compare with the reference case.

Same speed as the reference case.

Conceptual demonstration, not a prediction. Stokes' law assumes dilute, non-interacting spheres in a Newtonian liquid. Real products are crowded and often have a yield stress; if the yield stress is high enough, droplets may not move at all. Use it to understand direction and scale — doubling the radius quadruples the speed — then confirm with real stability data.

Emulsion

Physical structure
Droplets of one liquid dispersed in another liquid it does not mix with — oil in water (O/W) or water in oil (W/O). A film of emulsifier at each droplet surface resists the droplets merging. The continuous phase is often thickened.
Ingredient functions
Oil phase (emollients, oils, waxes); water phase; emulsifiers and co-emulsifiers; thickeners; preservatives; actives.
Stability mechanisms
Emulsions are thermodynamically unstable; the formulator slows the routes to failure: creaming or sedimentation (slowed by small droplets, small density difference and a viscous or structured continuous phase), flocculation, coalescence, Ostwald ripening and phase inversion.
Manufacturing variables
Temperature of each phase; order of addition; homogenisation shear and time; cooling rate; temperature at which heat-sensitive ingredients are added; deaeration. Emulsions depend on their process history.
Common failure modes
Visible separation; viscosity drift; grainy texture from crystallising fatty materials; phase inversion; preservation failure in the water phase.
Appropriate analytical methods
Microscopy and droplet size; viscosity; pH (for O/W); conductivity to confirm emulsion type; centrifugation and temperature cycling as risk screens; real-time stability.
Packaging considerations
Oils and emulsifiers interacting with plastics; shear in pumps; headspace oxygen; compatibility with metal tubes.

Limits of this tool. General descriptions and schematic diagrams, not formulation instructions. The settling bench applies Stokes’ law qualitatively and cannot predict the stability of a real product.

Conceptual demonstration

Experiment planner

Build a two-level factorial experiment: choose variables to change, variables to hold constant, a response and an acceptance criterion. The planner generates the runs in random order and, once results are entered, estimates each effect.

Many candidate variables, two levels each. The aim is to find the few that matter. A fractional design saves runs, at the price of confounding some effects with interactions.

Why structure matters

Prototype 2 used more emulsifier and a faster mixer than prototype 1, and its viscosity rose. Which change caused it? From those two batches, nobody can say: the two effects are completely confounded. A factorial design changes the variables in a balanced pattern so each effect can be estimated separately, and randomised run order stops drift in the lab (warming equipment, ageing materials) from masquerading as an effect.

1. Independent variables (choose 2 to 4)
2. Controlled variables

Not controlled: Same operator or a written procedure; Same measurement method and temperature; Recorded ambient temperature and humidity. Record them as nuisance variables, or their influence will be mistaken for an effect.

3. Response and acceptance criterion

Not yet set. An experiment without a predefined criterion invites moving the goalposts.

Full factorial: every main effect and interaction can be estimated separately.

Run orderDesign rowA: Emulsifier levelC: Homogenisation speedD: Cooling rateViscosity at defined conditions
14+ Higher level+ Faster− Slow cooling
28+ Higher level+ Faster+ Fast cooling
35− Lower level− Slower+ Fast cooling
43− Lower level+ Faster− Slow cooling
56+ Higher level− Slower+ Fast cooling
62+ Higher level− Slower− Slow cooling
77− Lower level+ Faster+ Fast cooling
81− Lower level− Slower− Slow cooling

Enter a result for every run (or use the fictional example) to calculate effects.

Limits of this tool. This is a genuine design-of-experiments calculation for two-level designs, but a simplified one: no replication, no error estimate and no significance testing. Example results are fictional and say so.

Conceptual demonstration

Failure investigation

Seven realistic formulation failures. For each, observe the symptom, propose causes, choose tests that tell them apart, read the results, reach a conclusion, choose a corrective action and decide how to verify it.

  1. 1 Observe the symptom
  2. 2 Select possible causes
  3. 3 Choose discriminating tests
  4. 4 Review the results
  5. 5 Reach a defensible conclusion
  6. 6 Propose corrective action
  7. 7 Decide how to verify

Step 1 of 7 · Observe the symptom

An oil-in-water face cream shows a thin oily layer at the top after four weeks in a heat-stress cabinet. Room-temperature samples of the same batch look normal. Two earlier batches passed the same screen.

First action, before any investigation: contain. Hold potentially affected batches so nothing is released while the cause is unknown.

ILLUSTRATIVE. Fictional products, fictional results. The scenarios teach a method of investigation; real investigations depend on the specific formula, process and evidence.

Limits of this tool. Fictional products and illustrative results. Each scenario teaches a way of reasoning; a real investigation depends on the actual formula, process and evidence, and no single solution fits every formulation.

Working templates

Download the templates

Each file includes instructions, field definitions and example entries. Example entries are marked ILLUSTRATIVE and describe fictional products. All ten templates are listed on the library page.

References and further reading

Sources

Primary sources for verification. Listing a source is not a claim that every requirement in it applies to every product. Recheck laws, standards and market dates before commercial use.

  1. Statutory regulation

    Regulation (EC) No 1223/2009 on cosmetic products (recast) (opens in a new tab)

    European Union, EUR-Lex. Scope: Cosmetic products made available on the EU market. Consolidated versions incorporate later amendments; check the version in force on your market date.

    Requires a responsible person in the EU, a safety assessment and cosmetic product safety report before placing on the market, a Product Information File, labelling and notification.

  2. Regulatory guidance

    Commission Implementing Decision 2013/674/EU — Guidelines on Annex I (cosmetic product safety report) (opens in a new tab)

    European Commission, EUR-Lex. Scope: EU cosmetic product safety reports.

    Guidance to responsible persons; it does not replace the qualified safety assessor, who alone carries out the safety assessment.

  3. Official regulator information

    Questions and Answers: FDA's regulatory actions on over-the-counter sunscreen (opens in a new tab)

    U.S. Food and Drug Administration. Scope: Sunscreens marketed in the United States.

    Sunscreens are regulated as over-the-counter drugs under monograph M020. In 2026 FDA added bemotrizinol as a permitted active ingredient and removed PABA and trolamine salicylate.

  4. Pharmaceutical methodology

    ICH Quality Guidelines: Q8(R2) Pharmaceutical Development, Q9(R1) Quality Risk Management, Q10 Pharmaceutical Quality System (opens in a new tab)

    International Council for Harmonisation. Scope: Medicines. Useful as scientific and risk-management methodology; not a legal requirement for cosmetics.

    Q8, Q9 and Q10 are designed to work together to support science- and risk-based approaches to quality.

Links and summaries last checked 18 November 2026.

Editorial methodology

These frameworks were developed and curated by Shamim Quresh, drawing on her professional practice in product development, formulation, manufacturing, branding and compliance.

They rely on published standards, regulations and established scientific principles, cited where they are used. That research and those publications are the work of their own authors and publishers.

AI tools assisted with drafting, structuring and checking the text, the interactive tools and the templates. Content is reviewed by Shamim before publication, and regulatory references were checked against official sources on the review date shown.

Worked examples, scenarios, results and data are illustrative. They do not describe experiments, tests or studies that Shamim has carried out, and no regulatory assessment of any product has been made.

Want to apply a framework to your own product?

I use these methods in consultancy work on product development, commercialisation, manufacturing and compliance. Tell me what you are building and where it is stuck.