Abstract
Small and medium-sized enterprises increasingly rely on public review platforms as signals of quality, reputation, and customer experience, yet many owners still lack a practical way to process multilingual review text at regular intervals. This study develops a low-cost and reproducible workflow for collecting Google Business reviews, filtering translated English content, and summarizing sentiment with open analytical tools. The pipeline uses SerpApi to retrieve recent reviews for a known Google Maps place identifier and VADER to classify translated review text into positive, neutral, and negative sentiment. The procedure is demonstrated on 200 Google reviews for an auto-rental location, Klass Wagen Sibiu. Results indicate a strongly positive corpus, with 85% positive, 4% neutral, and 11% negative reviews, a mean VADER compound score of 0.538, and a median of 0.733. The association between VADER scores and star ratings is high, r = 0.83, but it is interpreted as convergent descriptive evidence rather than classification accuracy. The study contributes an applied decision-support blueprint that connects online-review analytics, sentiment analysis, and SME monitoring practice. Its limits remain important: review data are self-selected, platform access is constrained, automatic translation can alter sentiment, and lexicon-based models may miss sarcasm or domain-specific wording.