How we score
One number built from what thousands of buyers already said, and exactly how we blend it.
The Glow Score
The Glow Score puts a candle on a scale from 0 to 10, with one decimal. It blends two kinds of buyer ratings: the candle's own rating on Amazon, and the Trustpilot score of the company that makes it. Both are out of 5, so we take their weighted average and double it.
Example shown.
What each source rates
- Amazon rates the candle. We use the average star rating and review count of the candle's listing on Amazon, sold by the brand itself where we can find it.
- Trustpilot rates the company. Its reviews cover shipping, packaging and customer service as much as the candles, so it says more about buying from a brand than about one scent. We use the TrustScore, or the plain average where Trustpilot shows no TrustScore.
Weights and thresholds
- Candles: Glow Score = 2 × (75% of the Amazon rating + 25% of the company's Trustpilot score). The candle itself carries most of the weight.
- Brands: Glow Score = 2 × the average of the Trustpilot score and the brand's Amazon average, weighted equally. The Amazon average covers every candle listing we matched, weighted by its number of reviews.
- Enough reviews to count: an Amazon rating counts from 10 reviews for a candle and 25 reviews in total for a brand; a Trustpilot score counts from 25 reviews. When only one source has enough, the score uses that one alone.
- No score without the candle's own rating: a candle gets a Glow Score only when its own Amazon rating counts. A company's Trustpilot score says nothing about how one scent smells or burns, so on its own it is not enough. Those candles show no score at all rather than a guess.
Bands
Every score comes with a word: 9.0 to 10 Exceptional · 8.0 to 8.9 Great · 7.0 to 7.9 Good · 5.0 to 6.9 Mixed · below 5 Poor. The number and the word always appear together, so the score never relies on color alone.
Checked dates
Ratings change, so we record the date we last checked every Amazon listing and Trustpilot page, refresh them regularly and recompute scores when we do. Brands can't pay to change a score, and claimed pages get no ranking advantage.
AI summaries
Where enough buyer reviews exist, we read them and publish a short summary of what buyers like and dislike, in our own words. Every summary is labelled “AI summary” with its source, how many reviews it covers and the date. We never republish review text. When there are too few reviews to say anything useful, we say so.
Scent families and notes
Notes are the ones each brand publishes. We file every candle in up to four of ten scent families, using the brand's own category when it has one and its notes otherwise. Spotted a mistake? Tell us and we'll fix it.