K-Brand Love Index
How do Indonesian students feel about Korean brands — and why?
The index measures emotional attachment and brand preference — feelings anyone can share, whether or not they have used the product.
K-Brand Love Ranking
Ranked by KBLI — emotional attachment (60%) and brand preference (40%). Experience, loyalty and advocacy are reported as separate modules and are never folded into the score. Scores show reliability grade and sample size.
Collecting brand love data
KBLI scores appear once a brand has 10+ approved, analyzed comments with evidence on both core axes. Tell us how a Korean brand makes you feel to help build the index.
KBLI is a reference brand-affinity index based on participating Indonesian university students. The published score covers emotional attachment and brand preference only, because product usage in this population is still limited; experience, loyalty and advocacy appear as separate modules once they have enough evidence of their own. It is not official national statistics representing all Indonesian consumers, and AI never infers a dimension the text does not support.
Evaluation Method
Overall Score (100 points)
The overall score is the average AI sentiment of approved one-line comments (0–100). No star ratings or tag weights are used — only the sentiment AI reads from the comment text itself.
Rising K-Brand
K-POP artists are evaluated on a separate index with music-specific axes, independent from the K-Brand 30 (K-Beauty / K-Food) score.
K-Brand Love Index (KBLI)
KBLI is a brand-affinity index, separate from the Sentiment Score above. Product usage among Indonesian students is still limited, so asking about purchase or recommendation would produce answers most respondents cannot honestly give. The published index therefore covers only what anyone can report: how a brand makes them feel, and whether they prefer it to the alternatives.
Evidence modules
Brand Experience, Purchase & Loyalty and Advocacy are still measured, but they are reported beside the index rather than inside it. A module appears only when it has at least 10 supporting comments and covers at least 30% of that brand's analyzed comments. A dimension with no evidence is shown as absent, never as a score of zero.
Where the evidence comes from
The site collects student opinion through two forms: the one-line brand comment and the Rising response. Both feed KBLI when they are about a consumer brand, so a student's answer counts the same wherever it was written. A Rising response reaches KBLI only when an administrator has linked that candidate to a K-Beauty or K-Food brand; K-POP candidates carry no such link and stay entirely within the Rising methodology.
Publication conditions
- At least 10 approved, AI-analyzed comments
- Evidence present on both core axes
- Average AI confidence of 70 or higher
- Administrator approval before a score becomes public
KBLI covers K-Beauty and K-Food brands only. K-POP artists keep the separate Rising K-Brand methodology and never receive a KBLI score.
Reliability Adjustments
- Prevent over-ranking of brands with few comments (Bayesian correction)
- Remove duplicate or near-duplicate comments (hash-based)
- Exclude or separately flag promotional, sponsored, or spam comments
- Automatically remove personal information before analysis
- Prioritize comments from the most recent 12 months
- Limit any single source to no more than 60% of the total score
- Reduce the weight of sarcastic or ironic expressions
Minimum Requirements for Official Ranking
- At least 10 approved comments
- At least 10 AI-analyzed comments
- Average AI confidence score of 70 or higher
- Relevance score of 60 or higher
- Brand information confirmed by admin
If these conditions are not met, the message "There is currently insufficient analysis data, so this brand is not included in the official ranking." is shown. If fewer than 30 brands qualify, only the actual number of eligible brands is announced.
AI Sentiment & Customer Experience Analysis
Both Korean and Indonesian expressions are analyzed. Cosmetics are structured and analyzed by quality, texture, effectiveness, scent, irritation, etc.; food products by taste, freshness, halal suitability, packaging convenience, etc. Mixed expressions (e.g., "The texture is good but the scent is disappointing") and promotional or sarcastic expressions are distinguished. Safety issues (allergies, foreign matter, etc.) are not published automatically and are forwarded to the admin.
Data Operation Principles
Only actually registered brands and actually written comments are analyzed. We do not generate fake brands, comments, scores, or rankings; when data is insufficient, we honestly show "Collecting comments". Comments containing personal information are automatically de-identified.