Download for iOS info@sift-scanner.com

Sift  /  Methodology

Transparency

Our principles and methodology.

How we score products, and why independence matters.

91

Sift Core Index

How scoring works.

Sift uses natural language processing to determine sentiment. We don't simply count keywords.

Our algorithm assigns weight based on context, source credibility, and consensus. A highly upvoted, detailed analysis carries more weight than a short, generic comment.

Scores reflect a composite of sentiment, ingredient quality, value for money, and reported experiences across multiple sources.

Weighting

Not every voice
counts the same.

Detailed, highly upvoted analysis Carries most
Considered comment with specifics Carries more
Short, generic comment Carries least
Detected spam or promotion Removed

Core principles

  • No sponsored placements
  • No affiliate links or kickbacks
  • No brand partnerships affecting scores
  • Independent data aggregation throughout
Close view of a Sift findings list. Two red warnings about talc and asbestos litigation, one amber note about fragrance allergens, and two green findings praising a reformulation to cornstarch and the removal of parabens and sulfates.
Findings are colour coded by severity. Good and bad sit in the same list, on the same product.

Platform limitations

Where Sift stops.

Not every product has enough data to form a score. If a product is too new or too niche, Sift will indicate insufficient data rather than guessing.

Sentiment is also a snapshot in time. It can change as products are updated, reformulated, or as new information emerges.