Glossary

Personalization Weight

Updated February 28, 2026

Personalization weight is the importance assigned to your safety profile, goals, adherence, and symptom history during ranking.

Why it matters

The more complete your profile, the more your recommendations reflect your actual constraints than generic population averages.

How weights are built

Think of them as adjustable priors:

These inputs combine with evidence metadata and recency weighting.

Why user inputs dominate

A single repeated behavior change can outweigh older sparse history because it changes current state:

Practical effect examples

Practical action step

Use a meaningful update window (3+ days) after behavior change before expecting ranking to settle.

Uncertainty and limits

Cross-site references

How this appears in Unfair

Personalization weights show up as why-same-ingredient recommendations rise, fall, or pause when your behavior changes between cycles.

Clinical safety note

Behavior changes can be dramatic; if significant symptoms emerge, prioritize safety settings over optimization weights.

Related

Feedback Loop

Your feedback loop is the process where your logs become the model input that reshapes your next recommendations.

Recommendation Confidence

Recommendation confidence is a probability-style indicator of how reliable current [ranking signals](/blog/complete-guide-to-supplement-stacks) are.

Recommendation Engine

The recommendation engine is the path from your inputs to [ranked suggestions](/blog/complete-guide-to-supplement-stacks), through filters and guardrails.