How does Pikkizy’s sentiment score work?
Pikkizy ranks products by combining two independent signals: YouTube creator review sentiment and live multi-retailer pricing. We analyse the transcripts of independent video reviews — the source video is public, so anyone can check what was said, though we do not verify creators’ identities — and extract aspect-level mentions across effectiveness, value, quality, and experience. The overall score is a weighted ratio of positive mentions, normalised across creators and refreshed every 30 days. A product needs at least 3 independent reviews from distinct channels before it earns a rating; below that threshold we publish no score rather than a thinly-evidenced one. Prices are scraped from multiple European retailers every 6 hours, and the price history graph shows the lowest, median, and highest price over the trailing 90 days. Affiliate partnerships never influence product rankings.
The five steps behind every score
- Discover— we identify YouTube reviewers covering the product’s category by channel authority and topical relevance.
- Transcribe — we use automated transcription on the public video stream and structure each transcript into timestamped segments.
- Classify aspects — an NLP pipeline labels each segment as positive, neutral, or negative for the four moat aspects: efficacy, value, quality, experience.
- Aggregate — we compute a weighted ratio of positive mentions, downweighting reviews that lack independent testing or read like sponsored content.
- Combine with price— the final ranking blends the sentiment score with current price competitiveness across the retailers we monitor for that product’s SKU.
Why this differs from retailer star ratings
EU Omnibus Directive Article 7 requires platforms to verify the authenticity of consumer reviews. Retailer star ratings can be incentivised, gamed, or skewed by review-for-discount programmes. Creator transcripts are auditable: the source video is public, so anyone can check what was said. We do not verify creators’ identities or credentials. Where retailer ratings still appear on Pikkizy product pages, they are clearly labelled “Aggregated retailer reviews” in the structured-data reviewAspectfield per Google’s review-snippet policy.
Affiliate independence
Pikkizy earns a commission when users click through to a retailer and complete a purchase. This commission never influences product rankings, scores, or editorial coverage — products are ranked solely by the sentiment + price algorithm above. Affiliate links carry the rel=“sponsored”attribute as required by Google’s linking guidelines and are visually disclosed above-the-fold on every product, best-of, and listing page. Read the full editorial standards for the complete policy.
Frequently asked questions
How is the sentiment score calculated?
We extract aspect-level mentions from the transcripts of independent YouTube creator reviews. Each mention is classified as positive, neutral, or negative for a given aspect (effectiveness, value, quality, experience). The overall score is a weighted ratio of positive mentions, normalised across creators and refreshed every 30 days.
Why don't you use retailer star ratings?
Retailer star ratings can be incentivised, gamed, or skewed by review-for-discount programmes. We rely on creator transcripts because the source video is public: anyone can watch it and check what was actually said. To be clear about what we do not do — we do not verify creators' identities, their stated credentials, or whether they bought the product, and we do not present these as verified consumer reviews.
How many reviews do you need before you publish a score?
A minimum of 3 independent reviews from distinct YouTube channels. Below this threshold a product appears with no rating to avoid thinly-evidenced rich results — Google's review-snippet policy treats those as a quality issue.
Do affiliate partnerships influence rankings?
No. Product ordering is computed solely from sentiment score and price competitiveness. Retailers cannot pay to improve a product's position. The affiliate disclosure on every product page is a legal Trust signal, not a sponsorship marker.
How fresh are the prices?
Multi-retailer prices are scraped on a 6-hour cycle. The price history graph on every product page shows the lowest, median, and highest price over the trailing 90 days, sourced from the same scrapers.