We recently hosted a webinar that delves deep into very topic of ChatGPT-based Text Analysis with Survalyzer. Starting with an enlightening introduction to ChatGPT’s capabilities with Survalyzer, the discussion transitions into DPD Switzerland’s customer-centric journey. Dive into our comprehensive webinar recording where Christian Hyka and Marco Kaiser explain the current state of text analysis.
Timestamp: [00:00:00]
The host of the event, Christian Hyka introduces the meeting agenda, explains its various sections, and explains the webinar’s topic, namely using ChatGPT for sentiment analysis and topic analysis of open-ended feedback responses.
Timestamp: [00:03:55]
Marco Kaiser, the Head of Marketing at DPD Switzerland, begins by discussing their transformative journey from a negative NPS trend to a customer-centric focus. Afterwards, he highlights the integration of Survalyzer for efficient feedback collection and analysis.
Furthermore, he underscores the company’s dedication to understanding feedback origins, setting clear objectives, and focusing on key customer touchpoints. Additionally, Marco emphasizes the importance of real-time feedback and champions the use of NPS as a pivotal metric for DPD.
Timestamp: [00:17:16]
Chris demonstrates the ideal approach to assessing open-ended responses. Using several examples of feedback to the question, “Please justify your purchase,” he elucidates the correct evaluation method. He delves into the different sentiments texts can be categorized into: positive, neutral, and negative.
Furthermore, Chris introduces another dimension: a list of topics used to analyze these comments. He then showcases a table where comments are aligned with specific topics and sentiment analysis.
Timestamp: [00:19:18]
Christian discussed the hurdles of traditional survey methods during the ChatGPT and Survalyzer webinar. He emphasized the tedious task of sifting through and categorizing every single response. Sometimes, it’s even tricky to tell if a comment is positive or just neutral.
He also touched on the challenges of multilingual surveys. Translating each feedback into a language everyone understands is time-consuming. Plus, there’s the risk of accidentally sharing personal details, like when a comment names a specific employee.
But there’s hope. Christian hinted at a solution that combines the strengths of ChatGPT and Survalyzer. This could be the answer to faster and more efficient survey analyses.
Timestamp: [00:22:39]
Before the rise of AI, software evaluation had its constraints. Chris, using Survalyzer Professional Analytics, first introduces us to a traditional survey dashboard. Here, users can easily set filters to focus on specific target groups. Next, he displays a table filled with open comments, which are straightforwardly listed. As he transitions, he highlights the word cloud, a popular tool that emphasizes commonly used words. However, he is labelling these methods as the “Old World” techniques, hinting at their outdated nature and limited automation capabilities.
Timestamp: [00:24:06]
Contradictory to the traditional approach Christian Hyka introduced new solution, emphasizing the integration of AI Sentiment and Topic Analysis. With this, users can utilize ChatGPT for survey analysis, streamlining the process through a guided setup. Users can choose to analyze all or specific survey responses, anonymize texts, and even translate comments. Notably, the business version leverages the ChatGPT API, which incurs a fee. After configuration, users can initiate the analysis, with processing time varying based on the volume of responses.
Timestamp: [00:27:35]
Christian Hyka unveiled Survalyzer’s “Open Comments” dashboard, which is harnessing AI for sentiment and topic analysis. The ChatGPT integration efficiently categorizes feedback into sentiments like negative, neutral, or positive, and highlights the frequency of discussed topics. For instance, while “simplicity” received positive nods, areas like “competitions” faced criticism.
Christian also emphasized the AI’s adeptness in translating comments, drawing parallels with platforms like DeepL. While AI streamlines the process, Survalyzer ensures quality by allowing manual edits. Users can tweak raw data, reassign comments, or even export for external adjustments, ensuring both automation and accuracy.
