August 22, 2026
Our Search Listing Gave One Rejection the Wrong Score
Our search listing for the Tensira throw rejection gave its final score as 7.6. The score we recorded when we turned it down was 7.80.
When we turn a product down we publish the record: what it was, what it scored, and which of our gates it failed. One of those records has been carrying two different scores, and only one of them was ours.
The product is a handwoven indigo throw from Tensira. We researched it in April, scored it, and could not publish it. The record explaining that has been live since.
The page itself never gave a final score. The search listing for it did, and the figure it gave was lower than the one we recorded.
What was wrong
The title and description we wrote for search results both said our score was 7.6. Our evaluation register, written on the day we made the decision, records 7.80. The record’s own summary line, on the page, says 7.8.
So a person who met this product in a search result read one number, and a person who opened the page read another.
Which one is right
The page scores four dimensions and weights them: WOW at 35 percent, GLOW at 25, MOMENT at 25 and TRUST at 15. This throw scored 8 on the first three and 6.5 on trust, and the record says so in full. Those four numbers weight to 7.775, which is what the register rounded to 7.80 and what the summary line rounds to 7.8.
7.6 cannot be produced from the scores the record itself publishes. It was typed by hand into a search field, once, and nothing has ever checked that field against the decision behind it.
What changed
Both search fields on that record now read 7.8. Nothing about the verdict moves: the product was below our threshold at either figure, and it stays turned down for the reason it always had.
How we found it
We were rebuilding the archive that lists our rejections, and before designing it we checked every score that page would print against the score each record stores. That check reads the hand-typed search fields too. Across all of our rejections, 122 records publish a score in a search field; 121 of them agree with what we recorded, and this one did not.
We would rather publish the one than quietly correct it.