Your purchase data is a product you're currently giving away for free, mostly to platforms.
You watch what tens of thousands of households in your category actually buy, week over week, with the returns attached. Ulta runs the paid version of this at scale: UB Media turns loyalty and purchase insight into a revenue line. The mid-market version isn't different in kind, just in sample size, and it's available to anyone willing to build the pipeline.
What you sell is aggregate, never personal. Category velocity. Size curve drift by region. Which two products keep landing in the same basket. What sold through in 6 days in March and 19 days in July. Your suppliers want it because they're forecasting blind. Your investors want it because it's a leading indicator. Some of your competitors want it too, and that's a pricing question rather than a moral one. Metafields are queryable in ShopifyQL now, so the analytics side is a pipeline job rather than a warehouse project, and a gated B2B storefront handles subscriptions, access, and billing.
The catch is that aggregation is not anonymization. Slice a small dataset finely enough and you can reconstruct an individual buyer, which is a real risk and a real liability. You need suppression thresholds, a privacy policy that genuinely permits secondary use, and a named owner for the review. The second catch is less legal and more brutal: a data product without a publishing cadence dies in 2 quarters. If nobody owns the report, you've sold a subscription to something that stops arriving.
First-party data has been framed as a defense for 5 years, a way to survive the loss of the pixel. It can also just be a line item that makes money on its own.
