How to Use K-Beauty Atlas — and Read the Graph Critically

K-Beauty Atlas is a visual index, not a compatibility score. Use it to move between ingredients, products, brands, concerns, and trend waves—then read each connection according to what the data actually represents.

Choose an entry point before opening the whole map

• Known ingredient: search its common or INCI name, then open the ingredient page.
• Known product: search the exact name or use the directory to inspect its reviewed formula.
• Skin goal: begin with a concern guide, then apply the same concern filter to the graph.
• Routine question: use Routine Builder rather than treating graph proximity as a layering rule.

Starting with a question keeps the dense graph useful. The full map is best for discovery after you know whether you are looking for a product, ingredient family, formula pattern, or historical trend.

1. Read the node before the line

Colored ingredient nodes are grouped by broad cosmetic function. Smaller neutral product nodes connect to ingredients mapped from their available ingredient lists. Node size and prominence help exploration; they are not measures of clinical effectiveness or concentration.

A function color is an editorial discovery category, not a complete biochemical identity. One ingredient can have several formulation roles, and its presence does not show the amount, delivery, stability, or performance of the finished formula.

2. Distinguish two kinds of connection

• Contains: a product-to-ingredient line means the ingredient was mapped to that product. It does not reveal an exact percentage.
• Interaction: an ingredient-to-ingredient line summarizes a layering consideration recorded in the Atlas, with confidence and source context where available.

3. Use focus before navigation

On desktop, hovering a node fades unrelated items and emphasizes its immediate connections. Clicking a node opens its full detail page. On a touch device, the first tap focuses the node and reveals its neighborhood; a second tap on the focused node opens the detail page. This preserves an exploratory step when hover is unavailable.

Focus shows a local neighborhood, not a ranking. A nearby or highly connected node is not automatically more suitable, effective, popular, or important for your skin.

4. Inspect evidence and confidence

Open an ingredient or hover a connection to see its type, note, confidence, and source link. An interaction candidate is excluded from the public graph until its source is reviewed for direct relevance and its wording is narrowed to that evidence. “Formula companions” only means two ingredients appear together in mapped products; it is not proof of synergy or safety.

• Formula-supported means the pair appears in directly relevant finished-formula or regimen evidence. It is deliberately narrower than claiming universal synergy.
• Confidence describes the editorial strength of that specific interaction note.
• Source lets you inspect what population, formula, concentration, or outcome was actually studied.
• Formula companion is a co-occurrence signal and should not be promoted into a causal claim.

5. Narrow the graph before zooming

Concern, wave, category, and brand filters reduce visual density and create a more useful starting point. Search is better for a known name. Use the full map to discover nearby products, then switch to detail pages when you need source and formula context.

Concern filters are discovery labels, not diagnoses. Trend waves describe the Atlas's editorial history model, not a product launch-date guarantee. Brand and category filters constrain what is shown; they do not change the evidence behind an edge. The URL records supported filters so a specific view can be shared and reopened.

6. Read detail pages in the right order

1. Confirm the exact ingredient or product identity and any aliases.
2. Check the reviewed market, formula status, source, and review date.
3. Separate declared percentages from ingredient-list position.
4. Read limitations before turning a cosmetic function into a result expectation.
5. Use linked products or formula companions as research leads, not endorsements.

Ingredient order is useful but limited. Under United States cosmetic-label rules, ingredients are generally declared in descending order, while ingredients at one percent or less can appear without strict predominance order. Atlas therefore displays position as label order, never as an invented concentration.

7. Turn discovery into a decision

• Compare two mapped product formulas side by side.
• Build a routine to place products in AM or PM and review duplicated ingredients.
• Browse concern guides for a smaller, goal-based entry point.
• Use the directory when you want an alphabetical index instead of a graph.

Comparison highlights shared and different mapped ingredients; it does not choose a winner. Routine Builder keeps AM and PM roles visible and surfaces sourced layering notes, but it cannot predict compatibility or replace product directions.

8. Avoid five common graph-reading errors

• Do not read node size as effectiveness or ingredient concentration.
• Do not read a short line as stronger evidence than a long line.
• Do not treat co-occurrence as synergy, safety, or a recommendation.
• Do not assume an absent product or ingredient is absent from the market.
• Do not merge regional or reformulated ingredient lists because the product name looks unchanged.

9. Keep the data limits visible

Product formulas change, regional lists can differ, ingredient order is not an exact concentration, and the current database is curated rather than exhaustive. The live data policy and coverage page reports dataset size, formula-link depth, interaction-source coverage, and the rules used to publish each edge.

A missing or incorrect record should be treated as a correction opportunity, not silently filled with an assumption. The methodology page explains source priority, update cadence, and how to submit a correction with the exact package, market, and supporting URL.