Make your next R&D decision clearer.

A useful evidence map begins with one exact sentence and the decision behind it.

A paper can be relevant to a subject without supporting every statement made about that subject. Before adding references, write the proposed wording, identify its audience and explain where it would appear. A technical discussion with a formulator and a statement on a consumer product may raise different questions.

Include the impression created by nearby images or comparisons. Then ask a practical question: what would a reasonable reader believe has been demonstrated? That is the proposition your evidence needs to address.

An ingredient name is a starting point. The preparation, formulation and batch make the record useful.

Record the subject of the evidence and the subject of the proposed statement separately. Is one a crude preparation and the other a purified fraction? Has a material been incorporated into a different vehicle? Are the tested and proposed concentrations actually known?

Use identifiers that make those differences visible. If the relationship is uncertain, leave it uncertain. A shared name should not fill a missing preparation history. The European Commission also makes a separate regulatory point: an INCI entry in CosIng does not itself establish that an ingredient is permitted or approved for use.

• Tested preparation and batch, where available.
• Proposed ingredient or formulation and intended conditions of use.
• Known changes in composition, processing or concentration.
• Unknowns that prevent a meaningful match.

The strength of an appraisal depends partly on what was available to examine.

Keep the citation or record reference together with a precise location, such as a figure, table or methods section. State whether you inspected the full text and supporting material, an abstract, a supplier summary or no underlying record.

For original work, link the conclusion to its underlying records in your controlled files. For published work, preserve the authors’ reported conditions and limitations. If the material is a hypothetical example, mark it as such at the start. None of these categories can silently substitute for another.

Write down where the observation was made before deciding where it applies.

Record the model or test surface, preparation state, exposure conditions, comparator and measurement time. Then place the proposed use alongside it. This makes the transfer question visible instead of hiding it inside a broad word such as performance.

For a hydration question, an inert film, a laboratory skin model and a human-use assessment are different evidence contexts. A useful map asks which part of the intended outcome each context addresses and what bridge remains to be established. The worksheet flags a selected laboratory-to-human transfer for review; it does not determine whether that transfer is justified.

Capture the result closely enough that another reader can check it.

Record the endpoint, units, timing, comparison and uncertainty reported in the source. Identify what was independently prepared, sampled or observed. If an effect size, analysis detail or limitation is missing, write “not reported” rather than filling the space with an estimate.

Keep observation and interpretation in separate fields. “The instrument reading changed” is an observation only when the underlying record supports it. What caused that change and what it means for a product are further questions. Reporting guidance should fit the study type; CONSORT, for example, addresses randomised trials and is not a universal score for every experiment.

• What changed, relative to which comparator?
• How was the endpoint defined and measured?
• What uncertainty and limitations were reported?
• Which parts of the conclusion are interpretation rather than direct observation?

The important work often sits between the result and the final sentence.

Build a chain from the verified observation to the proposed interpretation, then to the intended material, use and wording. Label each link with the evidence or assumption it depends on. If the chain changes preparation, formulation, model or population, show that change.

This is a map of reasoning, not a hierarchy in which every study must climb the same ladder. The question is whether the particular evidence fits the particular statement. The 2017 technical document on cosmetic claims discusses this relationship as nonbinding working-group best practice; it is not an official Commission position or a product-specific legal decision.

A strong alternative explanation helps identify the comparison that matters.

Ask what else could produce the observed endpoint. In a film experiment, consider whether ordinary differences in deposited mass, thickness, solids, salts or handling could account for the result. In an assay, ask whether interference or the comparator choice could change its interpretation.

The aim is to specify a discriminating question. “There may be confounders” is too vague to guide work. “Would the difference remain when deposited mass is matched?” identifies a condition that a reviewed design could investigate. The appropriate controls and analysis still depend on the material and method.

Evidence and public wording need to be considered together.

A scientific mechanism discussed in a research context can take on a different meaning when used to promote a product. Record the intended use and market rather than treating wording as a final cosmetic edit.

The FDA explains that claims and other aspects of presentation can establish intended use. Its examples include claims about changing melanin production and regenerating cells. These are reasons to seek a product-specific US classification review, not a reason for this worksheet to classify a product automatically. GB, EU and other market decisions also require their own current context.

• Describe the use the audience is being invited to expect.
• Record whether the wording is internal, educational or promotional.
• Flag unresolved classification or claims questions for the relevant specialist.
• Keep scientific review and permission to publish as separate decisions.

Narrow wording makes the next research question easier to see.

Begin with the actual subject, system and endpoint. Preserve meaningful conditions and uncertainty. If the source only supports a material-level observation, do not write the missing product-level conclusion into the sentence.

The comparison below is fictional. It shows how to frame the information to retrieve; it does not report an experiment, establish a benefit or provide approved claim language.

The best next step addresses a consequential gap.

Turn the largest unresolved inference into a bounded work item. State the question, proposed measure, comparison and decision consequence. If a result could not change the next decision, reconsider whether it is the right work to prioritise.

Give the work item an accountable role and identify what must happen first: access to the source, material characterisation, method review, rights, facilities or a specialist assessment. Keep proposed thresholds and resources visibly provisional until justified. This avoids presenting a research plan as a completed milestone.

Follow one fictional record all the way to its unresolved decision.

Imagine a team considering the sentence “H-01 improves skin hydration”. The only available information is a summary describing a laboratory film test on an inert surface. The underlying methods and results have not been inspected. H-01 is a fictional code, and there are no measurements or Bio results in this example.

The useful outcome is a better decision record: obtain the underlying source, identify the tested preparation, appraise the measurement and only then decide whether a human-use question is worth pursuing. Loading the example into the worksheet preserves those gaps. It does not create evidence.

The worksheet organises your reasoning and identifies selected gaps in the information you enter.

Start blank or load the fictional example. You can leave fields unknown and still download the record. Prompts describe missing information or selected transfer questions; they are not an efficacy score, an evidence-quality rating or approval to publish. An absence of prompts does not establish that the reasoning is complete.

Entries are held in this browser session. Download a JSON copy for your own records or use the print view for a discussion. Keep confidential source documents and identity keys in your controlled files, and use non-identifying references here. Review downloaded files before sharing them.

• One record for one material, intended use and proposed statement.
• Review prompts alongside the original sources and professional judgement.
• Record the next decision and its owner.
• Download before leaving if you want to keep your work.