Market research is increasingly incorporating artificial intelligence (AI), particularly through synthetic data generation or synthetic sample boosting, to address structural constraints such as small sample sizes and rising fieldwork costs. Despite growing adoption, the empirical validity of AI-generated data relative to traditional market research data remains insufficiently examined. This study analyses whether and under which conditions synthetic data—used both as fully synthetic panels and as a sample-boosting extension of observed survey data—can support managerial decision-making in market research applications. In collaboration with a Swiss market research institute, we first analyse a pilot application in which AI models learn statistical regularities from observed survey data and generate additional synthetic observations, effectively increasing sample sizes by a factor of two to three. We then replicate the analysis using panel data from Swiss Brand Observer across multiple industries and research contexts to assess robustness and external validity. To evaluate equivalence between observed and augmented data, we implement a difference-in-differences design that compares changes in key brand metrics—such as awareness, consideration, and purchase intention—between the original samples and their synthetically boosted counterparts over time. Pre-liminary results indicate that synthetic sample boosting can reduce estimation variance and support exploratory analyses when empirical data are constrained. At the same time, systematic deviations arise in certain subgroups and more granular analyses, highlighting limits to methodological rigor. Overall, the findings suggest that AI-based data augmentation can complement conventional market research under clearly defined conditions, but does not substitute for primary data collection in general.