Turning Data into Money: What It Means and Which Options Exist

The hosts outline three routes: sell data externally, use it to improve internal decisions, or build a product around it. Andrey immediately challenges the assumption that selling requires a large dataset. Different buyers value different information: a model developer might need a small, unusual corpus, an advertiser needs audience insights, and a manufacturer wants market information. Storage volume therefore cannot replace an explanation of the buyer’s task. Alexander connects this to pricing: information that seems peripheral to its owner may matter greatly to another company’s product. For internal use, Andrey distinguishes operational management, choosing the next action, and discovering hypotheses. Data can help at each level, but the planning horizons call for different questions. After establishing this frame, the hosts spend most of the episode on external sales and reserve the fuller internal-decision discussion for a later conversation.

Nikolay asks the group to separate having data in a system from having rights to use it. Being able to read a table does not explain why the company collected the information or what it can pass on. The personal-data examples bring collection purposes and informed consent into the discussion. The hosts then consider a messaging-automation service: conversations pass through its infrastructure and may remain in logs, although the user’s relationship with the messenger does not by itself describe the entire processing chain. Cloud providers have their own suppliers, turning an apparently simple arrangement into a network of dependencies. The point is to expose the organizational work hidden behind a promise to export some data. Nikolay’s legal examples illustrate his concern about risk; the hosts do not examine a specific transaction’s documents or establish a universal permission to sell a category of information.

The next question concerns de-identification. Removing a name, email address, or another obvious identifier sounds like a clear technical solution. The hosts instead draw attention to sequences of actions. Event timing, routes, and behavioral patterns may connect with other sources and point back to a person. Alexander describes combining several datasets: each supplier sees its own fragment, while the buyer can assemble a larger picture. Nikolay emphasizes the tension between usefulness and disconnecting records from reality. Buyers want information because it describes the world; reducing detail may reduce both risk and value. This discussion does not establish that every transformation is useless. It explains why deleting a few columns cannot answer every question about subsequent use, and why the seller’s assumptions about what a recipient can infer need scrutiny.

Andrey proposes selling aggregated analytics for a defined purpose instead of raw records. He discusses Strava, where movement information can complement knowledge about urban infrastructure. Users of a sports application, however, are not the same population as all city residents, so useful observations do not automatically form a representative picture. Retail analytics for suppliers offers another example: buyers want category sales, geographic context, and comparisons with competitors. Here the hosts disagree. Nikolay and Alexander worry that reduced detail lowers the price while preparation and maintenance costs remain. Andrey argues that a reliable answer to a useful question can itself be a product. He highlights coverage, representativeness, and stable collection. If a metric rises because the observation network expanded rather than because the market changed, a technically correct table can still support a poor decision.

The discussion then turns to preserving economic value. A buyer can retain or resell a one-off copy, so possessing useful data does not guarantee recurring revenue. The hosts discuss exclusivity and sports statistics: customers need information delivered at a particular speed, and access arrangements become part of the offering. The conversation moves from a file toward a product that must be maintained. Alexander contrasts two sellers. For a specialist analytics company, gathering and selling information is the main business. For a bank or another operating company, data may primarily support an advantage in its own product. External revenue must then be weighed against what the company gives away by sharing that knowledge. Buyers face a corresponding calculation: acquiring information makes sense when using it creates sufficient value, rather than merely making their archive larger.

Andrey includes preparation, maintenance, legal and reputational risks, and opportunity costs in the final calculation. Alexander distinguishes building a system from scratch from extending an existing platform. A large company may find another dataset technically straightforward, yet approvals can take long enough for the opportunity to disappear. Andrey notes the other side of scale: more sources bring more inconsistencies and duplicates. Nikolay adds that reconciliation skills are easier to reproduce where they already exist than to establish in a company without them. The group does not offer a general verdict that selling is worthwhile; a concrete use case, operating capability, and calculation are needed. They close by framing the next conversation around internal decisions. A report becomes meaningful when connected to an action that changes after someone reads it. Linking observation, decision, and financial outcome continues their central question about profit.