August 6, 2026
The Number We Rounded For Clicks
A theme helper moved the true product count before publishing it in category titles and search descriptions, while the grid and the structured data on the same page carried the real one.
A category page on this site tells you how many gifts it holds. Until today that number was not the real one. A helper in our theme took the true count and moved it before printing it, because a piece of search research says odd numbers get clicked slightly more often than round ones. A shelf holding 34 gifts published 31. Above sixty it moved the other way, so a shelf holding 61 published 67.
Nothing was broken. It did exactly what it had been built to do, which is why it was so easy to walk past.
What made it visible was reading one file all the way through instead of jumping to the part that needed changing. That file carries two other helpers whose entire purpose is that we do not round. One renders a price of 12.95 as 12.95 and never as 13. The other renders a score of 8.75 as 8.75 and never as 8.8. Both exist because a published price has to match the store exactly. Sitting in the same file was a third helper whose only job was to publish a number that was not the number.
The page had also been disagreeing with itself. The grid of products you see, the line naming how many were reviewed, and the structured data we hand to search engines have always carried the true count. Only the browser tab and the description that appears in search results carried the moved one. One address, three surfaces, two different answers, and no way for a reader to know which to believe.
The rounding is now deleted. Category, occasion, relationship, vibe and price pages publish their real count wherever they publish a count at all.
One rule is kept, and it is worth naming because it resembles the thing we just removed and is not the same thing. A shelf holding fewer than eight gifts still publishes no number at all. Choosing not to mention a figure is a decision about what to say. Saying thirty one when the answer is thirty four is a statement of fact, and it was a false one.
What we are taking from this: a number treated as approximate because it was only marketing is still a number we published under our own name. The rule that should have caught it had been sitting a few hundred lines up in the same file the entire time.