A consulting firm is preparing data for an AI-driven customer segmentation model.

Before any data preparation or modeling, PMI-CP–style guidance on AI initiatives emphasizes data quality assessment as the first critical activity. Quality must be evaluated before cleaning, enrichment, or labeling so that the team clearly understands the condition of the raw data and the scope of remediation needed. One of the primary quality dimensions to check early is completeness—whether required fields are present, whether key attributes are missing, and whether coverage is sufficient across the population of customers for meaningful segmentation.

If completeness issues are severe, downstream activities such as data cleaning, enhancement, and modeling may propagate bias or produce unstable segments. By systematically assessing data completeness first, the project manager enables the team to: (1) quantify gaps, (2) decide whether to obtain additional data, and (3) prioritize subsequent cleaning and enrichment steps. Data enhancement (option B) and cleaning (option C) are important, but they are remedial actions that should be guided by the initial quality assessment. Data labeling (option D) is more relevant for supervised learning use cases than for unsupervised customer segmentation. Therefore, to verify data quality prior to preparation, the project manager should first assess data completeness.