Modelling the interaction between a provider of composite bundles and a consumer who learns about her preferences or product quality through consumption, we show how the provider can strategically manipulate bundles to bias the consumer's learning. This can favor or disfavor certain goods, delay learning, and boost profits by leveraging temporary consumer biases. Using concepts from OLS estimation, decision theory, and network theory, we provide insights into obfuscation strategies and the role of the first and last eigenvectors in policy design under uncertainty. A toy empirical example from the movie industry illustrates the model's key mechanisms.
