A Comparative Evaluation of Data Analytics and Visualization Platforms: An Empirical Study of Power BI and Google Data Studio

Abstract

Business Intelligence, or BI, tools have become a main channel for turning massive amounts of business data into meaningful, decision-ready, and visual insights. Microsoft Power BI and Google Data Studio lie among the most widely adopted self-service platforms. However, limited empirical research comparing them under the same implementation circumstances is available. This study introduces an implementation-based evaluation framework in which analytical objectives, source datasets, dashboard specifications, ETL transformation goals, and scoring criteria are held constant while each platform follows its own preprocessing route: Power Query for Power BI and Excel-based preparation for Google Data Studio. Four functionally equivalent dashboards are constructed to ensure visual outputs remain directly comparable. The platform, together with its data-preparation environment forms the independent variable. Ten evaluation dimensions—ETL capability, dataset handling, observed interaction effort, KPI monitoring, filter integration, chart quality, geographic visualization, dashboard consistency, analytical capability, and no-code suitability—are scored on a five-point rubric by five independent evaluators, using an Amazon retail dataset (50,000 records) and a Walmart retail dataset (1,000 records) as controlled inputs. Power BI recorded a mean total of 40.2 out of 50 and Google Data Studio 32.4 out of 50. Power BI scored higher on ETL flexibility, workflow reproducibility, DAX-driven analytical modelling, and automatic label spacing in dense charts. Google Data Studio required fewer interaction steps during dashboard construction, offering faster customization, and produced detailed geographic maps through its native Google Maps engine. The study highlighted that dashboard outcomes were associated with distributional profile of the underlying data: near-uniform distributions tended to yield monitoring-oriented overviews, while more variable distributions highlighted distinguishable analytical patterns. These findings offer empirically grounded and scenario-based guidance to assist practitioners in selecting a platform that suits their analytical requirements in no-code BI environments, with retail analytics serving as the empirical context for this evaluation.