What sells together, and what's just coincidence
Market basket analysis identifies items that appear together more often than chance would predict. Golden ranks pairs by lift and shows when the data is too thin to support a bundle.
Your top pair is probably your two best sellers
Count which items show up together and the winner is almost always the two most popular things in the store. Jumbo bags go with jumbo bags because everybody buys jumbo bags.
Counting tells you what's popular. Bundles, cross-sells, and shelf placement need to know what's connected, which is a different question with different math.
Rank by surprise, then check the math
Golden ranks pairs by lift, how many times more often two items land together than chance would predict, so popularity stops drowning out real connections.
Every pairing has to clear a statistical bar before it shows up, groups of items that travel together get called out as bundles, and when your baskets can't support a conclusion, Golden says so.
How it works
Golden spots the baskets
Any table shaped like a group and its items works: orders and products, respondents and answers, users and features. Golden needs at least 20 baskets that hold more than one thing.
It measures lift
For every pair: support (how often both show up), confidence in each direction, and lift (how many times more often they appear together than if they had nothing to do with each other). "Wooden Heart and Wooden Star land in the same basket 30x more often than chance."
It checks the math
A pair needs at least 1.5x lift, and the 95% confidence interval on that lift has to stay above 1. Golden gives you the pessimistic read too: "Even on a pessimistic reading the pairing holds at 23x."
It finds the bundles
When three items all pair with each other, Golden calls out the group, so you can treat them as one decision instead of three.
It tells you when there's nothing there
Small baskets spread the evidence thin. When no pairing beats chance once Golden accounts for how many it tested, the board says so in one sentence at the top. We'd rather tell you that than hand you a bundle built on luck.
What you get
- 01A ranked pairs table plus a common-vs-surprising scatter of every pairing
- 02Confidence in both directions, because a rare item brings a common one along far more often than the reverse
- 03Click a pair to filter the board to the baskets that hold it
- 04Works for any basket: carts, survey answers, feature usage
A bad bundle costs real money
A bundle built on coincidence eats margin and shelf space. A real one shows up in the data months before anyone notices it in the store. Golden finds the real ones and is honest about the rest.
Find out what travels together
Sign up for early access and run a basket analysis on your own orders.