Stars tell you how customers rated a product. Sentiment tells you how they felt when they wrote about it. When QI reads the words, a 4 star review that reads like a complaint counts as negative, so problems show before your average moves.
Where: Analytics ▸ What customers say ▸ Sentiment, and the full summary on product pages
Covers: Your whole store in Analytics, one product on its page
Plans: Growth and up
The Sentiment card
➀ Where it comes from. QI analysis of N reviews across your store, or Based on star ratings in this period.
➁ The split. The share of positive reviews in the middle, and each group with its count and share.
QI analysis of N reviews across your store. The numbers come from: QI's reading of the words in reviews, added up across every product that has a What shoppers say summary. It covers those reviews whatever the dates you picked.
Based on star ratings in this period. The numbers come from: The stars of the reviews published in your dates: 4 or 5 stars count as positive, 3 as neutral, 1 or 2 as negative. Shown until your products have summaries.
QI reads a product's reviews when it writes its What shoppers say summary, the first time a shopper opens the product page with 5 or more reviews. So the QI figure grows as shoppers visit your products.
On your product pages
The full What shoppers say summary on each product page has a sentiment bar for that product's reviews, with the share of positive, neutral and negative.
How to act on it
Watch the negative share, not the positive one. A rise of a few points over a month is your earliest warning.
Find where it comes from. Open the products' summaries: the downsides in Top mentions say what's wrong, such as sizing or delivery.
Check after a change. New packaging, a new supplier or a new size chart: sentiment is your before and after.
Replying to the negative reviews helps too: shoppers read your replies, and a fixed problem said out loud wins trust back.
💡 Tip: When stars and sentiment disagree, trust the words. Read the reviews behind a high rating with a low sentiment: they're polite customers telling you what to fix.