Data Quality and Bias in AI Marketing Models: Sources, Consequences, and Mitigation Strategies

Abstract

This article reviews how artificial intelligence transforms marketing analytics while remaining constrained by data quality and algorithmic bias. Drawing on a systematic literature review of peer reviewed studies, the paper synthesises evidence on how artificial intelligence supports segmentation, personalised targeting, dynamic pricing, and automated service when embedded in integrated data and model architectures. The findings show that information quality, consistency, and governance across data pipelines are preconditions for reliable prediction, because missing, noisy, or poorly integrated data distort customer classification and managerial insight. The review also reveals that historical, sampling, measurement, and feedback loop biases in training data and modelling choices can generate systematically unfavourable outcomes for vulnerable customer groups, eroding trust, brand equity, and regulatory compliance. Overall, the study argues that accurate and fair artificial intelligence marketing models require socio technical solutions that combine robust data governance, transparent model monitoring, and organisational accountability for outcomes across customer subgroups. The article concludes with an agenda for future interdisciplinary research directions.