How we score products
isitbuzz measures the gap between what a product's marketing claims and what real owners actually report. Every score is deterministic math over auditable evidence — the same inputs always produce the same score. We never change a score for money.
1. Where the data comes from
For each product we collect opinions and claims from multiple independent source types, in parallel:
- Reddit — real user discussions in relevant communities
- YouTube — video metadata, comments, and whether the reviewer is sponsored
- Editorial reviews — professional review sites
- Forums — specialist community forums
- Manufacturer & retail listings — the official marketing claims
- Multilingual sources — non-English reviews for global coverage
Snippets are de-duplicated and re-ranked for relevance before anything is scored.
2. User Satisfaction (0–100)
Each opinion is classified by sentiment (positive / mixed / neutral / negative) and severity, and we detect whether the author is a verified owner. We then compute the share of meaningful opinions that are positive, weighting negatives by severity, giving extra weight to verified-owner experiences, and diluting for low-signal neutral comments.
The result is calibrated so that 50 means genuinely mixed — some love it, some don't. Small samples are pulled toward 50 because we're less certain. A high score means owners are consistently satisfied, not that the marketing is loud.
3. Marketing Hype (0–100)
We classify each marketing claim as vague, verifiable, or inflated, and check each one against real user feedback for contradictions. The hype score combines how often claims are contradicted by owners, how vague or inflated they are, and how many of the top YouTube reviews are sponsored. Higher means more misleading.
When we can't collect any marketing claims for a product, we show the hype and gap as unavailable rather than inventing a flattering number.
4. Trust Rating & Verdict (1.0–5.0)
The Trust Rating blends three things — real user experience (the largest factor by far), how honest the marketing is, and how strong the underlying evidence is — into a single 1.0–5.0 score. It maps to a plain-language verdict so you don't have to interpret a number:
- RECOMMENDED — users are consistently satisfied and the marketing holds up
- LIKELY SAFE — mostly positive, with minor caveats
- NEUTRAL — genuinely mixed; it depends on your priorities
- CAUTION — recurring problems, or marketing that overpromises
- AVOID — owners report serious, widespread issues
5. Confidence
Every analysis reports a confidence level based on how many opinions we found, how many came from verified owners, and how many independent source types contributed. Fewer sources and fewer owners mean lower confidence — and a more cautious score.
What we don't do
- We don't average star ratings — those are easy to game.
- We don't accept payment to change a score. Money never moves a number.
- We don't treat sponsored reviews as if they were independent.