You look up an ingredient in EWG Skin Deep and it comes back with a high hazard score. You check the same ingredient in another database and it’s rated low concern. A third source lists it as not well studied. Three different answers for the same ingredient, using overlapping research — and no obvious way to know which one to trust.
This kind of inconsistency is one of the more frustrating parts of ingredient research, and it’s common enough that it deserves a direct explanation. The databases aren’t necessarily wrong — they’re measuring different things, using different frameworks, and making different judgment calls about what the available evidence means. Understanding why they diverge makes all of them more useful.
How Ingredient Safety Databases Work
The first thing worth understanding about ingredient safety databases is what they actually are — and what they aren’t.
They’re not independent testing laboratories. EWG Skin Deep, CosDNA, INCIDecoder, and similar resources don’t conduct original research on ingredients. They’re curated databases that aggregate existing research — peer-reviewed studies, regulatory assessments, industry safety data, and government agency evaluations — and apply a scoring or classification system to summarize what that research shows.
What that means in practice is that a database’s rating is only as good as the research it’s drawing from, how complete that research is, how recently it was updated, and how the database’s methodology weighs different types of evidence. Two databases can look at the same ingredient, draw on largely the same body of research, and produce different ratings because they made different decisions about how to interpret and weigh what they found.
This isn’t necessarily a flaw in the system — it reflects the complexity of ingredient safety evaluation. It does mean that understanding what each database is actually measuring is important for using any of them accurately.
Why the Same Ingredient Gets Different Scores
Several factors explain why the same ingredient produces different ratings across different databases — and most of them come down to methodology rather than disagreement about the underlying facts.
Different hazard frameworks are the most fundamental source of divergence. Some databases rate ingredients based on the full range of potential hazards associated with a compound — what it’s been shown to do at some dose, in some context, or through some exposure route. Others rate ingredients based on a narrower set of concern categories that reflect their specific focus, such as skin sensitization or irritation. An ingredient that scores high in a broad hazard framework might score lower in a more targeted assessment because the specific concern categories being evaluated are different.
Different evidence weighting is another significant factor. Some databases may give greater weight to animal study findings when human data is limited, while others may apply a higher threshold when human evidence is unavailable. Sources may also differ in how they weigh industry-funded studies, independent research, regulatory assessments, and other types of evidence. The same underlying research can therefore contribute to different conclusions depending on the methodology used to evaluate it.
Different concentration thresholds can also affect how ingredients are evaluated. An ingredient may have documented effects at high doses while appearing at much lower concentrations in consumer products. A database focused primarily on hazard may flag the ingredient based on its demonstrated capacity to cause harm, while an assessment that incorporates typical exposure levels may reach a different conclusion about the practical risk.
Different update frequencies mean that databases aren’t always working from the same body of current research. An ingredient that was evaluated using the research available several years ago may look different when evaluated against newer findings. Different databases also update at different rates, meaning the research informing their assessments may not always be from the same period.
The Hazard vs. Risk Distinction
The most fundamental conceptual difference between database approaches — and one that can explain significant rating disagreements — is the distinction between hazard and risk.
Hazard refers to the inherent capacity of a substance to cause harm — what it’s capable of doing under some conditions, at some dose, through some exposure route. Hazard assessment asks: can this compound cause harm? Under the right circumstances, many substances can cause harm at sufficiently high doses. Hazard assessment does not necessarily account for whether those circumstances are likely or relevant to actual consumer product use.
Risk refers to the probability of harm occurring under specific, real-world conditions — accounting for factors such as dose, frequency of exposure, route of exposure, and the population being exposed. Risk assessment asks: is this compound likely to cause harm at the levels people actually encounter it, through the way they actually use products containing it?
A database that emphasizes hazard will flag a compound that has been shown to cause harm in certain studies or under certain exposure conditions. That information can be useful for understanding what a compound is capable of, even when the conditions of those studies differ from typical consumer exposure.
A database or regulatory agency that evaluates risk may reach a different conclusion when the typical concentration in a consumer product and the typical exposure route do not correspond to the conditions under which harm has been documented.
These approaches are answering different questions. Hazard-based information helps describe what a substance is capable of doing under particular conditions. Risk-based assessments consider the likelihood of those effects occurring under specified exposure conditions. Because both approaches can appear as a simple score or designation on a database, it isn’t always obvious to the reader which question is being answered.
The Data Gap Problem
A significant proportion of ingredients in use in consumer products haven’t been studied well enough to produce confident safety conclusions either way — and different databases handle that uncertainty differently.
When EWG Skin Deep lists an ingredient as “fair” or notes data gaps, it can reflect insufficient research rather than evidence of either safety or harm. The ingredient hasn’t been studied enough to establish what it does, which is different from an ingredient that has been studied and found to be safe.
Some databases interpret data gaps more conservatively, while others treat the absence of evidence of harm as a reason for lower concern. This difference in how uncertainty is handled can produce significant rating differences because the databases aren’t necessarily disagreeing about what the research shows — they may be applying different approaches to what should be concluded when there isn’t enough research to reach a clear conclusion.
The important distinction is that “not enough data” is not the same as “safe.” It means the available evidence is limited or inconclusive.
Which Databases to Know and How to Use Each
Understanding what each commonly referenced database is designed to do makes it easier to use them for the right purposes. Not every database is evaluating ingredients in the same way, and some are designed more for education or product comparison than for assessing safety. The most useful tools fall into two broad categories — personal care and food — though some, such as Yuka, cover both.
Personal Care Databases
EWG Skin Deep is one of the most widely used consumer-facing ingredient databases for personal care products. It rates ingredients on a scale from 1 to 10 based primarily on hazard information — what available research suggests an ingredient is capable of doing across health endpoints such as carcinogenicity, endocrine disruption, developmental toxicity, and skin sensitization. It draws on scientific studies, regulatory assessments, and other sources. Its primary value is as a broad hazard-screening tool, particularly for identifying documented concerns and data gaps. Its limitation is that a hazard rating does not by itself indicate the level of risk associated with a specific concentration or exposure.
CosDNA focuses specifically on cosmetic and personal care ingredients, with particular attention to potential acne and irritation concerns. It provides ingredient information across safety, acne, and irritation categories using a simpler scoring system than EWG. Its primary value is for people researching how particular ingredients may relate to skin compatibility, rather than as a comprehensive assessment of every potential health endpoint.
INCIDecoder is primarily an ingredient education resource. It explains what individual personal care ingredients do in a formula, their common functions, and other information about their use in cosmetics. Rather than treating it as a definitive safety-rating system, it can be useful for understanding what an ingredient actually is and why it appears in a product before looking further into its safety profile.
SkinSort is a skincare-focused ingredient analysis and product comparison tool. It allows users to compare products, examine ingredient lists, and filter products according to particular ingredient preferences or concerns. Its strength is product-level comparison — for example, looking at how the formulations of two similar moisturizers or cleansers differ — rather than serving as a substitute for primary research on an individual ingredient.
Glowlens takes a formulation-focused approach to skincare ingredients, making it useful for looking beyond individual ingredients and considering how ingredients are combined within a particular product. This type of analysis can provide additional context when comparing formulations, although it should still be distinguished from regulatory or scientific safety assessments.
Osana is a personal care product-rating tool that uses barcode scanning to provide consumers with a quick assessment based on its own ingredient criteria. Its convenience makes it useful as an in-store reference, but as with other consumer rating tools, understanding the methodology behind its ratings is important before treating the score as a definitive safety assessment.
Yuka covers both personal care and food products. Users can scan product barcodes and receive a rating based on the criteria used by the platform. Its accessibility makes it useful for quick product comparisons, but its scoring methodology differs from databases such as EWG Skin Deep and CosDNA. A Yuka score therefore shouldn’t be treated as interchangeable with another database’s rating — the underlying criteria and weighting matter.
CosIng — the European Commission’s Cosmetic Ingredients database — is different from consumer-facing rating tools. It provides information on cosmetic ingredients within the European Union regulatory framework, including regulatory restrictions and references to relevant safety assessments. It can be particularly useful when researching how a cosmetic ingredient is treated under EU regulations and comparing that framework with the U.S. regulatory approach.
FDA resources can provide another perspective because they reflect the U.S. regulatory framework rather than a consumer-oriented safety score. FDA resources can help readers verify regulatory information and understand how ingredients are treated under U.S. law. For food ingredients, the FDA’s resources include information such as the Substances Added to Food inventory. For cosmetics, FDA resources provide information about applicable regulations, prohibited or restricted ingredients, and other regulatory requirements. These resources are most useful for understanding the regulatory baseline rather than obtaining a single safety score.
Food Databases
Open Food Facts is an open-source food product database containing ingredient lists, nutritional information, additives, and other product information from around the world. It is particularly useful for looking up packaged foods, comparing ingredient lists, and identifying additives found in specific products. Because the database relies heavily on community contributions, coverage and accuracy can vary by product and region, so information should be checked against the product’s current packaging when possible.
Bobby Approved is a consumer-facing food product evaluation tool that uses its own ingredient criteria to assess packaged foods. Its barcode-scanning format makes it convenient for quick in-store product checks. Like other rating systems, its results reflect the standards built into the platform, so understanding what ingredients it screens for — and what it does not evaluate — is important when interpreting its ratings.
Yuka also covers food products, allowing users to scan packaged foods and receive a product score based on the platform’s methodology. In addition to nutritional factors, its food ratings incorporate information about ingredients and additives. Because Yuka combines multiple factors into a single score, it can be useful for product comparisons while still requiring an understanding of how that score is calculated.
The FDA’s food ingredient resources provide a different type of information from consumer-facing apps. Rather than assigning products a score, FDA resources document the U.S. regulatory position on substances used in food. These resources can help establish whether a particular substance has a recognized regulatory status in the United States and provide a useful point of comparison when consumer databases flag the same ingredient for other reasons.
How to Use Multiple Databases Together
Once you understand what each database is designed to do, the next step is knowing how to use them together. No single resource provides every piece of information you may need, so the most useful approach is to treat databases as complementary sources rather than competing authorities.
Start by identifying the question you’re actually trying to answer. Are you trying to understand what an ingredient does, identify potential hazards, compare two products, check how an ingredient is regulated, or find the underlying research? The answer determines which resource is most useful.
For an unfamiliar ingredient, you might begin with an ingredient education database to establish what the substance is and what function it serves in a product. From there, a hazard-focused database can help identify documented areas of concern or gaps in the available evidence. If the ingredient raises questions about regulation, a regulatory database can provide information about how it is treated under a particular legal framework. When you want to go beyond database summaries, the original scientific literature and regulatory documents provide the underlying evidence.
The same approach works at the product level. A barcode-scanning or product comparison tool can help you quickly identify products or compare formulations, while the actual ingredient list allows you to verify what is currently in the product. If a particular ingredient stands out, you can then research that ingredient separately rather than relying solely on the product’s overall score.
When different sources give you different results, go back to the methodology. Ask what each source is measuring, what types of evidence it considers, whether it is evaluating hazard or risk, and whether the assessment accounts for factors such as concentration and route of exposure. Differences become much easier to interpret once you know what question each source is answering.
A practical research process might look like this:
Identify the ingredient or product → understand what the ingredient does → check for documented concerns → look at the regulatory position → review the underlying evidence when needed.
This approach also helps prevent a common mistake: treating a database score as the conclusion rather than the beginning of the research process. A rating can point you toward a question, but understanding the reasoning and evidence behind it provides the context needed to interpret that rating.
The databases covered above can make ingredient research considerably more accessible. Used together — and with their limitations in mind — they provide different pieces of the same larger picture.
Tools, Not Oracles
Ingredient databases are research aggregators — they make existing scientific and regulatory information more accessible to people who aren’t toxicologists or regulatory scientists. That’s genuinely valuable. What they aren’t is a single authoritative answer that resolves every question about every ingredient.
Understanding what each database is measuring, why ratings sometimes diverge, and what data gaps actually mean transforms these tools from a source of confusion into a more useful foundation for ingredient research. The goal isn’t certainty — it’s a more informed starting point for understanding what goes into and onto the body.
The references used in this article are a starting point — we encourage you to read further and draw your own conclusions.
New to ingredient awareness? Browse our starter guides for practical next steps across every category.
Leave a Reply