How K-Beauty Oracle Scores Products

What the scoring engine does

K-Beauty Oracle scores 155 products from 65 Korean beauty brands against your skin profile. The engine matches ingredient research to your concerns, weighs clinical evidence, checks texture compatibility with your skin type, and accounts for your experience level. No brand pays to be included or ranked higher. Every product enters the same scoring pipeline.

Evidence tiers

Each product's hero ingredient is assigned an evidence tier based on the depth of published research behind it. The tier determines a multiplier applied to the product's score. Ingredients with stronger clinical backing get a higher multiplier; traditional remedies with limited formal study get a lower one.

How scoring works

When you complete the diagnostic quiz, the engine scores every product in the catalog against your profile. The formula combines five factors into a single number per product.

• -Concern matching: Your primary concern counts 3x, secondary concern 1.5x, and tertiary concern 0.75x. Each product has pre-computed concern scores (0-10) for conditions like acne, hyperpigmentation, and dehydration. A serum with a concern score of 8 for acne and your primary concern is acne contributes 8 * 3 = 24 points.
• -Concern duration: Chronic concerns get a 1.15x multiplier (longer-term issues benefit from stronger actives). Temporary concerns get 0.95x.
• -Texture compatibility: Products are classified as light, medium, or heavy texture. Oily skin gets +2 for light textures and -2 for heavy ones. Dry skin is the reverse: +2 for heavy, -1 for light. This pushes gel moisturizers toward oily skin and rich creams toward dry skin.
• -Evidence tier multiplier: The concern + texture subtotal is multiplied by the hero ingredient's evidence tier (1.4x down to 0.85x). A product with a Proven-tier ingredient scores 65% higher than one with a Traditional-tier ingredient, all else being equal.
• -Potency tiebreaker: A small fraction (potency / 100) breaks ties between products that score identically on the above factors.

Additional scoring adjustments

The engine applies targeted adjustments for specific skin situations that need special handling.

• -Sensitive skin penalty: Products are rated on intensity (1-10). For sensitive skin, anything above intensity 3 gets penalized: intensity 4-5 scores at 60%, intensity 6-7 at 30%, and intensity 8+ at just 10% of its base score.
• -Unavailable products: Products flagged as currently unavailable on Amazon score at 30% of their normal value. They can still appear as alternatives but won't be the top pick.
• -Oily + dehydration combo: Light textures (emulsions, essences) get a +3 bonus. Heavy textures (creams, occlusives) get a 0.5x penalty. Hyaluronic acid products get +2. This prevents the engine from recommending thick creams to oily skin that just needs water-binding hydration.
• -Sensitive + acne combo: High-intensity BHA and salicylic acid products get a 0.2x penalty. Centella/cica and azelaic acid products get +3 each. The engine also caps your experience level at intermediate, even if you selected advanced, to avoid recommending harsh actives.
• -Hero ingredient deduplication: Within each routine step, only the top-scoring product per hero ingredient is kept. This prevents your serum slot from being filled with four different niacinamide options.

Experience gating

Products tagged as advanced-level (high-concentration retinol, strong chemical peels) are filtered out for beginner users. Intermediate and beginner users with chronic concerns can still see advanced products, but with a warning flag suggesting gradual introduction. If you have sensitive skin and acne, the engine caps your effective experience level at intermediate regardless of what you selected.

What we don't do

The scoring engine has clear boundaries on what it is and is not.

• -We don't test products on skin. Scores are based on ingredient research and product data, not our personal use.
• -No brand deals or sponsorships. Products enter the catalog through automated sourcing pipelines that search Korean beauty databases and Amazon listings.
• -No sponsored placements. There is no mechanism for a brand to pay for a higher score or preferred position.
• -No pay-to-rank. The scoring formula is the same for a $8 cleanser and a $45 serum.
• -Recommendations come from published ingredient research, product formulation data, and the concern-matching algorithm described above.

Keeping products current

The catalog refreshes weekly. Every Monday, an automated pipeline searches for new Korean beauty products on Amazon, classifies them using ingredient analysis, validates images, and merges them into the catalog. On Wednesdays, a comparison pipeline checks whether new products outperform existing ones for specific concern categories and flags potential swaps for review. A separate availability sweep runs twice weekly to mark products that have gone out of stock on Amazon.

Common questions

Does K-Beauty Oracle accept paid product placements?

No. Every product enters the catalog through the same automated sourcing pipeline and is scored by the same formula. There is no way for a brand to pay for inclusion or a higher ranking. The engine treats a $8 toner and a $40 serum identically in the scoring math.

How often does the product catalog get updated?

The catalog refreshes on a weekly schedule. New product searches run on Mondays, catalog comparisons run on Wednesdays, and availability checks run twice a week. Products that go out of stock on Amazon are automatically penalized in scoring so they drop out of top recommendations.

Why does a product with a Traditional-tier ingredient sometimes still appear in my routine?

Evidence tier is one factor, not the only one. A Traditional-tier product (0.85x multiplier) can still outscore a Proven-tier product if it has very high concern scores for your specific skin issues and good texture compatibility with your skin type. The multiplier difference between Proven and Traditional is significant (1.4x vs 0.85x), but a strong concern match can overcome it.

What data does the scoring engine use about each product?

Each product has concern scores (0-10 for each skin concern), a hero ingredient with an evidence tier, a texture classification (light/medium/heavy), potency rating, intensity rating (1-10), experience level tag, price tier, and time-slot preference (morning, evening, or both). These fields are populated during the sourcing pipeline using ingredient analysis.

Can I see why a specific product was recommended for my routine?

Yes. Each product in your routine shows a reasoning line that explains why it was picked, referencing your primary concern and the product's hero ingredient. You can also expand any product slot to see alternative products that scored well for that step, along with options at different price tiers.