FREE STANDARD SHIPPING $75+ 🇨🇦 FREE STANDARD SHIPPING $75+ 🇨🇦

Why Your Ingredient-Scanning App Is Getting the Science Wrong

Why Your Ingredient-Scanning App Is Getting the Science Wrong

Here is what an ingredient list for an apple would look like.

Yes, the kind you pick from a tree. If you had to put it on a label, this is what it would read:

Water, Fructose, Glucose, Sucrose, Malic Acid, Citric Acid, Ascorbic Acid, Dietary Fiber (Cellulose, Hemicellulose, Pectin, Lignin), Glutamic Acid, Aspartic Acid, Leucine, Alanine, Valine, Serine, Glycine, Threonine, Proline, Isoleucine, Phenylalanine, Lysine, Histidine, Arginine, Tryptophan, Methionine, Cysteine, Tyrosine, Palmitic Acid, Oleic Acid, Linoleic Acid, Stearic Acid, Palmitoleic Acid, Alpha-Linolenic Acid, Potassium, Phosphorus, Calcium, Magnesium, Sodium, Iron, Zinc, Manganese, Copper, Selenium, Fluoride, Retinol, Thiamine, Riboflavin, Niacin, Pantothenic Acid, Pyridoxine, Folate, Tocopherol, Phylloquinone, Quercetin, Catechin, Chlorogenic Acid, Epicatechin, Phloridzin, Procyanidin B2, Cyanidin-3-Galactoside, Ethyl Acetate, Butanol, 2-Methylbutyl Acetate, Hexanal, Trans-2-Hexenal, Ethanol, Amyl Acetate.

That's 57 chemicals. In an apple. Including acids, alcohols, aldehydes, and a list of amino acids that would make a skincare label look terrifying if you didn't know it was lunch.

Run that through a "clean beauty" scanning app and watch it panic.

The Apps

You've probably heard of at least one. Maybe you use one. Maybe it's the reason you're reading this.

Yuka. Over 55 million users worldwide. You scan a barcode, it gives you a score from 0 to 100, with a colour: green, orange, or red. Green means safe. Red means danger. It takes about two seconds.

EWG Skin Deep. Run by the Environmental Working Group. It rates ingredients on a scale of 1 to 10. Low is good. High is bad. It's been around since 2004 and has become the default reference for the "clean beauty" movement.

Think Dirty. Similar idea. Scan, score, sort. Products get rated on a "dirty meter" from 0 to 10, where 10 is the dirtiest.

They all do roughly the same thing: take an ingredient list, run it through a database, and give you a number that tells you whether the product is "safe" or "toxic."

The concept is appealing. The execution is where it falls apart.

The Foundational Error

There's a distinction in toxicology that every first-year student learns. It's the single most important concept in evaluating whether a substance is safe, and it's the one these apps get wrong.

Hazard vs. risk.

A hazard is the ability of a substance to cause harm under any circumstances. Arsenic is hazardous. The sun is hazardous. A kitchen knife is hazardous. Water is hazardous. You can drown in it.

A risk is the likelihood that a substance will actually cause harm at the concentration and exposure you're getting.

Arsenic in your drinking water, at the trace levels present in municipal water supplies, is not a meaningful risk. The sun on a five-minute walk to your car is not a meaningful risk. A kitchen knife sitting in a drawer is not a meaningful risk.

The difference between the two is everything. It's the difference between "this substance can cause harm under extreme laboratory conditions" and "this substance will cause harm in the amount present in your face cream."

These apps score hazard. All of them. They look at whether an ingredient has ever been associated with a negative outcome, at any dose, under any conditions, and they assign a risk score based on that.

They don't account for concentration. They don't account for how the ingredient is delivered. They don't account for how long it stays on your skin. They don't account for the regulatory limits that exist specifically to keep cosmetic ingredients within safe margins.

A toxicologist from the 1500s named Paracelsus figured this out five hundred years ago. He said it in Latin, but the translation is simple: the dose makes the poison.

Everything is toxic in the wrong amount. Everything. Including water. Including the vitamins in that apple. The question is never "can this substance cause harm?" The question is "will it cause harm at this dose, in this product, on this skin?"

These apps don't ask the second question. They only ask the first. And then they give you a colour.

How the Scoring Actually Breaks

Let me show you how this plays out in practice.

Phenoxyethanol is a preservative. It's used in skincare at concentrations of 1% or less. It has an extensive safety record. Regulatory bodies in the EU, UK, and Australia have evaluated it and set maximum usage levels that include wide safety margins. Cosmetic chemists consider it one of the standard, well-understood preservatives in the industry.

Yuka rates it as "moderate risk." That orange dot means the best score your product can get, regardless of everything else in the formula, is 50 out of 100.

So a well-formulated moisturizer with ceramides, hyaluronic acid, and a clinically supported barrier-repair profile gets capped at "mediocre" because it contains a preservative that's been safely used for decades, at a concentration well within regulatory limits, and that exists in the formula specifically to stop the product from growing bacteria.

(Without that preservative, by the way, the product goes rancid. Bacteria festers. Mould grows. The "clean" version is the dangerous one.)

Meanwhile, a product with no preservative system, questionable stability, and a shelf life that's anyone's guess could score higher because it doesn't contain the ingredient the app decided to flag.

The worst-ingredient-wins problem. Yuka's scoring works by flagging the single worst-rated ingredient in a product and using that ingredient to cap the total score. One "moderate risk" ingredient, even at a tiny concentration, drags the entire product into orange territory. It doesn't matter what else is in there. It doesn't matter how the formula was designed. The weakest link, as the app defines it, sets the ceiling.

This is like rating a restaurant based on the single ingredient in the kitchen the reviewer found most concerning, regardless of how it was used, in what amount, or what the actual dish tasted like.

The data gap penalty. EWG's Skin Deep database does something even more questionable. When it doesn't have enough data on an ingredient, that lack of data counts against the score. No data doesn't mean the ingredient is dangerous. It means it hasn't been studied as extensively as, say, water. But in EWG's system, "we don't know much about this" scores the same as "this has known concerns."

That's backwards. The absence of evidence is not evidence of harm. But in these databases, it's treated that way.

Who's Behind the Curtain

This is the part that tends to surprise people.

These apps are not regulatory bodies. They're not scientific institutions. They're not nonprofits operating in the public interest.

They're businesses. With revenue models.

Yuka earns roughly $7 million a year, primarily from premium subscriptions ($10 to $50 per year) that give you access to product recommendations and additional features. Those recommendations are, essentially, "your product scored poorly, here's a better one." The worse your current product scores, the more useful the premium subscription feels.

(Think about that incentive structure for a moment.)

EWG runs the "EWG Verified" certification program. Brands pay a $500 non-refundable application fee just to be considered. Ongoing licensing fees range from hundreds to thousands of dollars per year, with a royalty on licensed product sales over a typical three-year agreement. Brands that display the EWG Verified mark on their packaging have paid for the privilege.

There's nothing inherently wrong with charging for certification. But when the same organization that assigns the safety scores also sells the "safe" badge, and the badge is the solution to the problem the scores created, that's a closed loop worth noticing.

Think Dirty similarly partners with brands and sells premium features. Same structure. Score the problem, sell the solution.

None of this means their data is fabricated. But it means their incentives aren't perfectly aligned with giving you the most accurate, nuanced assessment possible. Fear drives engagement. Engagement drives subscriptions. Subscriptions drive revenue.

A calm, contextual answer like "this ingredient is fine at this concentration" doesn't generate the same response as a red warning screen.

How Safety Actually Works

If the apps aren't the authority, who is?

The actual safety assessment of cosmetic ingredients is done by bodies you've probably never heard of, because they don't have apps and they don't have marketing budgets.

In North America, the Cosmetic Ingredient Review (CIR) is the independent expert panel that evaluates the safety of cosmetic ingredients. It's been operating for 50 years. Its panel includes toxicologists, dermatologists, and chemists who review the full body of scientific literature on each ingredient and determine safe usage levels based on concentration, exposure, and biological activity. Not hazard alone. Risk.

In the EU, the Scientific Committee on Consumer Safety (SCCS) does the same work. Its opinions form the basis for the EU's cosmetic regulations, which are among the strictest in the world.

These bodies evaluate ingredients the way a toxicologist would: what does the evidence show at real-world concentrations, with real-world exposure patterns, in real-world product types? That evaluation includes wide safety margins. When the SCCS sets a maximum concentration for a preservative, that limit already includes a buffer, typically 100x below the level at which any adverse effect was observed in testing.

That's how safety works. Not a colour on a screen. A body of evidence, evaluated by specialists, with built-in margins for error.

The apps bypass all of this. They take the ingredient name, check it against their own database, and assign a score that ignores everything the regulatory bodies considered when they approved that ingredient at that concentration in the first place.

The Parabens Problem

If you want to see the real-world damage this kind of scoring causes, look at parabens.

Parabens are preservatives. They've been used in cosmetics for over 80 years. They are among the most studied ingredients in cosmetic history. The CIR has reviewed them. The SCCS has reviewed them. Health Canada has reviewed them. The FDA has reviewed them. The consensus, across all of these bodies, is that parabens are safe at cosmetic concentrations.

One study in 2004 found trace amounts of parabens in breast cancer tissue. The study did not establish that parabens caused the cancer. It didn't even establish a correlation. It found that parabens were present, which isn't surprising, because parabens are present in a huge range of products that most people use daily.

But the apps flagged it. The media ran with it. "Parabens cause cancer" became a cultural fact despite being a scientific non-event. Brands scrambled to reformulate. "Paraben-free" became a marketing claim. And an entire category of safe, effective, well-understood preservatives was driven out of formulations by fear.

What replaced them? In many cases, preservative systems that are less stable, less studied, and less effective. Some replacements caused more sensitization reactions than the parabens they replaced. The apps didn't flag those, because the new preservatives didn't have decades of studies for the databases to misinterpret.

The parabens story is the template. Flag an ingredient. Strip it of context. Create fear. Watch the industry reformulate around the fear. The product gets worse. The app gets credit for "protecting" you.

The Failure Mode of Information Seeking

Here's what bothers me most about this whole topic.

Many of you consider yourselves information seekers. And you are. You research. You read labels. You compare. You care more than the average consumer about what goes on your skin, and I respect that enormously. It's one of the things I like most about this community.

But information seeking has a failure mode. And the failure mode is stopping at the most convenient source.

Maybe it's an app that scans barcodes and assigns safety scores based on criteria you've never questioned. Maybe it's an influencer who says "I only use clean products" with enough confidence that it sounds like expertise. Maybe it's an ad designed to look like a documentary. Maybe it's just the packaging. "Plant-based. Clean. Chemical-free." Printed in earth tones on recycled cardboard with a leaf on it.

Something about it resonates. Your search ends. And now you know the truth.

Except you don't. You know a story. And the story was specifically engineered to make your search stop right there.

Real information seeking is uncomfortable. It means sitting with nuance instead of certainty. It means accepting that natural doesn't mean safe and synthetic doesn't mean dangerous. It means understanding that "chemical" is not a dirty word. Water is a chemical. Oxygen is a chemical. Every ingredient in that apple is a chemical. The word just means "a substance with a defined molecular structure." That's it.

(A lead pipe is "chemical-free." It's also not something you'd rub on your face.)

The apps give you certainty. Science gives you probability, context, and conditional answers. Certainty feels better. It's also usually wrong.

What to Actually Look for Instead

I'm not going to tell you to throw the apps away. If they help you identify specific allergens you react to, they have some value for that narrow purpose. If you know you're sensitive to a particular fragrance compound or preservative, scanning a label before you buy saves time.

But for general "is this product safe," the apps are the wrong tool. Here's a better hierarchy:

Regulatory bodies first. If an ingredient is approved for cosmetic use by the SCCS, Health Canada, or the FDA at the concentration present in your product, it has passed a safety evaluation that accounts for dose, exposure, and biological activity. That's the baseline.

Peer-reviewed research second. Not a blog post. Not an app database. Not an Instagram infographic. Published research in toxicology, dermatology, or cosmetic science journals, where the methodology is transparent and the findings have been reviewed by other scientists.

Cosmetic chemists third. People who actually formulate products for a living understand how ingredients behave at real concentrations in real formulations. They know the difference between "this ingredient is hazardous in its pure form at industrial concentrations" and "this ingredient is safe at 0.5% in a wash-off product." The apps don't make that distinction. Formulators do.

Apps and influencers last. Not because they're always wrong, but because they lack the context to be reliably right.

Where We Stand

I'll be direct. We use ingredients that some of these apps will flag. Preservatives. Surfactants. Silicones. They're in the formulas because they serve a function, they're safe at the concentrations we use them, and removing them would make the products worse.

We're a small brand. We don't have the kind of reach required to make sweeping change in how people evaluate ingredient safety. We can't compete with an app that has 55 million users and a two-second scan.

What we can do is keep explaining. Keep citing the science. Keep being honest about what's in the products and why. And trust that the people in this community, who care enough to read a 3,000-word blog about ingredient scoring methodology, are the kind of people who want the full picture, not just the colour.

Context matters more than a score. Dose matters more than a name. And the question you should be asking isn't "is this ingredient on a list?" It's "is this ingredient safe at this concentration, in this product, on my skin?"

The answer, almost always, is yes. Because that's what the actual science, the regulatory science, the boring science with no app and no marketing budget, was designed to ensure.

 

 


Frequently Asked Questions

Are ingredient-scanning apps like Yuka accurate?

They're consistent, but consistency isn't the same as accuracy. The apps score based on ingredient hazard profiles pulled from databases, without accounting for concentration, formulation context, or exposure duration. A preservative at 0.5% in a rinse-off cleanser gets the same score as that preservative at 100% purity in a lab setting. The score is "correct" by the app's methodology. The methodology itself is the problem.

What is the EWG Skin Deep database?

It's a database run by the Environmental Working Group that rates cosmetic ingredients on a scale of 1 to 10 based on potential hazards. It has been criticized by toxicologists and cosmetic chemists for conflating hazard with risk, for counting data gaps as negative marks, and for operating a paid certification program (EWG Verified) alongside the database that assigns the scores.

What does "the dose makes the poison" mean?

It's a foundational principle in toxicology, attributed to Paracelsus in the 1500s. It means that any substance can be harmful at a high enough dose, and virtually no substance is harmful at a low enough dose. Water can kill you if you drink enough of it. Arsenic is present in your food at trace levels that cause no harm. The safety of any substance depends on the dose, not on the substance's existence.

Are parabens actually dangerous?

No, at cosmetic concentrations. Parabens have been reviewed by every major regulatory body in the world and found safe for cosmetic use at approved levels. The 2004 study that triggered the scare found parabens present in breast cancer tissue but did not establish causation. The CIR, SCCS, FDA, and Health Canada have all reaffirmed parabens' safety since then.

Why do these apps flag preservatives?

Because preservatives are biologically active substances, they have to be in order to prevent microbial growth. That biological activity shows up in hazard databases. But the question isn't whether a preservative is biologically active. It's whether it's active enough, at the concentration used, to harm your skin. The answer, within regulatory limits, is no. Without preservatives, products develop bacteria, mould, and fungi. The "cleaner" version is the more dangerous one.

Should I stop using ingredient-scanning apps?

Not necessarily. If you use them to check for specific allergens you know you react to, they're a useful shortcut. But for general product safety, they're unreliable. They oversimplify toxicology, ignore concentration, and create fear about ingredients that have been evaluated and approved by actual regulatory scientists. Use them as a starting point if you must, but don't treat the score as a verdict.

How do I know if a cosmetic product is actually safe?

If it's sold legally in your country, it has met the safety requirements of your national regulatory framework. In the EU, that framework is among the strictest in the world. In Canada and the US, the CIR and Health Canada evaluate ingredient safety at real-world concentrations. For additional assurance, look for peer-reviewed research on specific ingredients you're concerned about, or consult a cosmetic chemist. A barcode scan is not a safety assessment.

Do brands pay to get on "clean" lists?

In many cases, yes. EWG Verified charges brands a $500 application fee plus ongoing licensing fees that can range from hundreds to thousands of dollars per year. Leaping Bunny certification is paid. Various "clean beauty" retailer lists have partnership or listing fees. This doesn't automatically make the certifications meaningless, but it means the organizations assigning the scores have a financial relationship with the brands being scored.

 

 

 

 


Sources

  • Nohynek, G.J., et al. (2010). An assessment of the genotoxicity and human health risk of topical use of kojic acid. Food and Chemical Toxicology, 48(5), 1317-1326. DOI: 10.1016/j.fct.2010.02.034 (discusses hazard vs. risk framework in cosmetic safety)
  • Darbre, P.D., et al. (2004). Concentrations of parabens in human breast tumours. Journal of Applied Toxicology, 24(1), 5-13. DOI: 10.1002/jat.958 (the original study that triggered the paraben scare)
  • Cosmetic Ingredient Review Expert Panel. (2008). Final amended report on the safety assessment of methylparaben, ethylparaben, propylparaben, isopropylparaben, butylparaben, isobutylparaben, and benzylparaben as used in cosmetic products. International Journal of Toxicology, 27(Suppl 4), 1-82. DOI: 10.1080/10915810802548359
  • Scientific Committee on Consumer Safety, SCCS. (2013). Opinion on parabens: Updated request for a scientific opinion on propyl- and butylparaben. SCCS/1514/13.
  • Romanowski, P. (2024). Quoted in "As Yuka Gains Millions of Users, Beauty Industry Insiders Question the Veracity of Its Product Ratings." Beauty Independent.
  • Isaacs, D. (2024). Quoted in "The Rise of Ingredient Anxiety." Marie Claire UK. ("In science, the dose makes the poison.")
  • EWG. "How We Work With Companies." ewg.org/EWGVERIFIED/how-we-work-with-companies.php (EWG Verified fee structure)
  • American Council on Science and Health. "Dear EWG, Why Are You Shaking Down Companies to Get on Your 'Verified' List?" acsh.org (EWG Verified criticism)
  • Novakovich, J. "A Case Against the EWG." The Eco Well. theecowell.com/blog/a-case-against-the-ewg