Most data businesses are built around the part that can be sold off.
There is a kind of data business that does not look like a data business, its compounding cannot be sold off, and the dataset is the wrong place to look for value. It is the most powerful data business that can be built.
Wrap is what happens when data is not extracted, packaged and sold, but woven around an activity that already exists, and the activity becomes worth more for the wrapping. The data has no separate price. Removing it does not reduce the activity by some percentage; it changes what the activity is.
The shape of the activity varies. In transactions money changes hands and the wrap improves the unit economics of each instance. In experiences attention is held and the wrap deepens what the attention is worth. In decisions a professional chooses what to do and the wrap raises the quality of the choice. The architectural property is the same in all three. The commercial logic of each is not.
In a transactional business the wrap is most visible because the unit of value is countable. Every successful advertising platform built in the last twenty years is a wrap business: the placement is what gets charged, but the placement is only worth what the data layer makes it worth. The same holds in payments, where Stripe’s fraud and acceptance-rate data is built into the payment route rather than sold alongside it. The platform cannot quote a price for the wrap; it can only quote a price for the activity, and the wrap shows up as the difference between that price and what a competitor without it would have to charge.
Experiential wrap is harder to recognise because the activity is consumed rather than transacted. A Premier League broadcast today is not the broadcast of 2010. The video feed has not improved meaningfully. What has changed is the data layer — expected goals, motion tracking, win-probability — and the experience that data layer creates. None of it is sold to viewers. Its commercial impact appears at three layers: the rights value the league can charge, the CPMs the broadcaster can charge advertisers, and the retention rate of the streaming subscriber. None of those three markets is for the wrap itself. A new entrant cannot purchase it; it accumulates only across seasons of broadcasts.
Decision wrap appears in professional environments where people are paid for the quality of the choices they make. The Bloomberg Terminal is its purest expression. Bloomberg does not sell financial data; the terminal is the operating system of professional finance, and the data is woven into the workflows, the chat, the analytics, the order entry. The same pattern operates in industrial buying, where a category manager’s range and promo decisions are made inside a data layer their supplier has built around the relationship. A buyer in a retailer-brand joint business plan has no obvious alternative venue. The decisions are made where the data lives.
Across the three shapes the property is the same. The wrap accumulates with each instance of the underlying activity, and only the operator keeps the accumulation. Participants benefit from it while they remain in the activity; they cannot take it with them. It cannot be priced separately because it has no separate purchaser. It is captured downstream — at the price of the placement, the rights deal, the seat licence — never at the wrap layer itself. A competitor cannot acquire the wrap by buying the dataset; the dataset is the residue of the wrap, not the wrap. Replicating the wrap requires replicating the activity beneath it, which is normally infeasible.
The clean separation between routes does not survive contact with a real organisation. A serious data programme will usually be doing Improve and Wrap simultaneously. Stripe’s fraud signal lowers Stripe’s chargeback costs while being woven into the merchant’s payment product. dunnhumby’s data work at Tesco tightened internal category management while restructuring how brands negotiated with the retailer. The question is not which route a programme is on. The question is which route is the commercial centre of gravity. The test is where the compounding lives, where the moat is, and where the value is captured rather than absorbed. If the network effect is internal and the savings stay on the operator’s P&L, the centre is Improve. If the network effect spans participants in the activity and the value is converted into pricing power, retention or switching cost, the centre is Wrap. Programmes confused about which they are typically build neither route well.
The question senior teams should ask of their data is not what it could be sold for, but what activity it could be wrapped around and which of their existing activities are not yet wrapped. Most companies have answers to both questions and choose to look the other way, because wrap creates no line item, requires no new business unit, and rewards no individual within twelve months. Sell does all three. Sell wins the budget meeting; Wrap wins the marathon. Sell is a product, Wrap is an environment.
What cannot be sold cannot be matched.
Gianluca Carrera, May 2026.
If you found this useful, get the next essay in your inbox — every two or three weeks. Subscribe to the newsletter.