What can a trip across town, a purchase, or a web search reveal about you? Each is an action that can become data. When companies combine location, browsing, and transaction data, they can use it to understand behavior and target messages or offers. section 2.7
Human behavior is not fixed. Seasons, holidays, and economic conditions can change what people do. Predictive models use records of past actions to anticipate future ones, so the data must remain relevant to current behavior. Recent data is often more valuable, but it must also be accurate enough for the decision at hand. section 2.4.7
When past behavior stops predicting the future
During the COVID pandemic, businesses saw models that had worked in January 2020 become less reliable in February and substantially wrong by March. People had changed how they acted. Retraining a model might help, but an abrupt shift can require recalibration or a complete rebuild. section 2.4.7
Behavior also includes decisions about what to share. People may strongly value privacy yet act in ways that reduce it. That is not necessarily a simple contradiction: privacy needs change with circumstances, relationships, and the competing desire to be social. In one survey, 93% of adults said controlling who can get information about them was important, while 90% said the same about controlling what is collected. section 5.7 section 5.3
Commercial use of behavioral data has grown from studying how to sell household goods to targeting content, products, and services across a much wider data ecosystem. Your actions can help organizations make predictions, but you can also ask who collects those actions, how they are combined, and whether the resulting uses serve your interests. section 1.4.3 section 2.7
