Methodology
How we score, compare, and recommend products.
Data Collection
Product data is sourced from public retail listings and manufacturer specifications. We pull pricing, features, and availability from multiple retailers to build a comprehensive spec sheet for each product in a comparison.
Scoring
Every product is scored on a 0–100 scale using a deterministic algorithm. The score is computed from measurable attributes — specifications, pricing position, and feature completeness. The same product with the same data always receives the same score, regardless of when or where it appears on the site.
Scores are not subjective ratings or editorial opinions. They are computed outputs based on structured inputs. We do not round, adjust, or editorially override scores.
Editorial Picks
Each comparison designates up to three picks: Best Overall, Best Value, and Best Premium. These are assigned deterministically based on score and price position — not editorial preference.
- Best Overall — highest composite score across all attributes
- Best Value — highest score-to-price ratio
- Best Premium — highest score among products in the upper price tier
AI and the Trust Boundary
We use AI (large language models) to generate editorial prose: feature commentary, trade-off summaries, FAQ answers, and category-level insights. However, the AI operates within a strict trust boundary:
- The AI never chooses which product wins or loses
- All scores, prices, badges, and product recommendations come from deterministic data
- AI-generated text is validated to ensure it does not contradict the structured data
- Product names in AI commentary sections are injected after generation, not produced by the model
Freshness
Comparisons display a “last updated” date reflecting when the product data was last refreshed. Our automated pipeline periodically re-checks pricing and availability. If a product is discontinued or its data becomes stale, it is flagged for review.
What We Don't Do
- We do not accept payment for reviews or placement
- We do not manually adjust scores to favour any product
- We do not use affiliate links that could bias recommendations
- We do not aggregate user reviews or sentiment as a scoring input
Questions
If you have questions about how a specific comparison was scored, or want to report inaccurate data, contact us at hello@lensavi.com.