Salentein – Media Monitor
Monitoring customer feedback and perception is crucial when making strategic decisions. This is precisely the challenge that the renowned winery, Salentein, wanted to tackle. Salentein wanted to gain insights into how consumers perceive their wines by monitoring online platforms. By doing so, they aimed to identify strong and weak features of their products, improve their positioning, marketing messages, and overall customer experience.
To achieve this, we extracted over 92,000 reviews about Salentein wines from the popular online wine platform, Vivino. For each wine and vintage, we extracted the price and reviews, as well as the rating, review text, and home country of the user for each review. To ensure that language barriers did not hinder the analysis, we translated each review to English using the Google Translate API, which recognized the language of the review automatically.
Next, we used ChatGPT, a state-of-the-art language model trained by OpenAI, to analyze the review text. Specifically, we used ChatGPT to split the review into subsentences and identify if each subsentence was about appearance, scent, taste, price, setting, or food pairing. This allowed us to gain a more detailed understanding of the reviews and identify common themes that emerged across reviews.
After analyzing the reviews, we combined all subsentences for each unique combination of wine, vintage, country, and topic. We then used ChatGPT to summarize these themes, resulting in a Tableau dashboard that shows the number of reviews and average rating per wine, vintage, country, and topic. The dashboard also includes a summary and word cloud that can be filtered by any field, allowing Salentein to quickly identify trends and insights.
These insights could help Salentein improve their product offerings by focusing on the features that customers value the most. They can also use the results to enhance their marketing messages, positioning, and customer experience. Additionally, the tool enables Salentein to track trends in customer feedback over time, providing a valuable tool for monitoring customer perception of their brand.
Kevin HaverLead Data Scientist
Laurien FortuinData Scientist