How GiftingMachine ranks gifts

Most “best gifts” lists are someone's opinion dressed up as a ranking — or worse, an auction where the top spot goes to whoever paid for it. We wanted the opposite: a list that earns your trust because the order is set by data you could verify yourself.

The inputPeople swipe, not click

In the GiftingMachine app, people swipe through decks of gift ideas — right for “I'd give this,” left for “not for me.” It's a low-stakes, honest signal: nobody is performing for an audience or leaving a five-star review to be polite. Every swipe is one person quietly telling us whether a gift actually lands.

Those swipes are the only thing that feeds the rankings. We don't count page views, we don't count clicks on a buy button, and we don't let our own team override the order.

The mathWhy a simple average isn't enough

The obvious way to rank gifts is by their right-swipe percentage. But a raw average is easily fooled: a gift swiped right by 9 out of 9 people scores a perfect 100% — yet nine swipes tell us almost nothing. A gift swiped right by 4,800 out of 5,200 scores “only” 92%, but we're far more certain it's genuinely loved.

So instead of the raw rate, we use the Wilson lower bound — the bottom edge of a 95% confidence interval around the true approval rate. In plain terms: it asks “given how many people swiped, what's the lowest the real love-rate could plausibly be?” A small sample gets pulled down hard; a large, consistent one barely moves. The result rewards gifts that are both well-loved and well-tested.

The promiseWhat money can't buy

This is the part that matters most: there are no sponsored placements anywhere on GiftingMachine. A brand cannot pay to appear on the board, to rank higher, or to be added to a guide. The only way onto the list is to be loved by the people swiping.

FAQThe honest questions