Google Reviews Analysis with Claude: Instructions for Use
Online reviews are a valuable source of information for understanding customer opinions. Analyzing them strategically is essential for identifying areas for business improvement. In this tutorial, I will explain how to use the artificial intelligence Claude to perform sentiment analysis on the Google reviews of your business in a simple and effective way. By following this practical guide, you can gain useful insights to make data-driven decisions that increase customer satisfaction.
Why Analyze Sentiment with Claude
Online reviews are essential for understanding customer experience and satisfaction levels. However, manually analyzing them is very time-consuming.
That’s why relying on Claude to automate sentiment analysis is the best choice for entrepreneurs. Claude is an AI developed by Anthropic that can quickly process large volumes of reviews and provide valuable insights.
With Claude, you can rapidly understand if the overall customer sentiment is positive, negative, or neutral. Additionally, Claude can automatically identify the most discussed topics in the reviews.
By leveraging Claude, you can obtain strategic information quickly to improve customer satisfaction and boost business performance.
Retrieving Google Reviews
The first step is to obtain the raw data from Google. One solution is to use online tools like Outscraper.com that automatically extract reviews and metadata.
Alternatively, you can web scrape the Google pages with the help of Chrome plugins, as in the case of this tutorial where we will use the extension called Instant Data Scraper.
Let’s see the step-by-step process:
- Open the Google reviews page of interest and sort them by most recent.
- Once done, launch the Instant Data Scraper plugin. Generally, the plugin automatically recognizes the various reviews (as in the image), but if this doesn’t happen, click on “Try another Table” until you get the desired result.
- Once everything is set, we can scrape the reviews by clicking on “Start Crawling” and saving the file in .csv format, as it is readable by Claude.
Data Cleaning
After downloading the raw reviews, it’s necessary to prepare the data for processing with Claude.
In our case, this simply means opening the .csv file with a spreadsheet like OpenCalc and deleting all columns except the “dehysf” column and the “Jtu6Td” column, which contain the date and the actual review text, respectively.
Analyzing Sentiment
Once the final file with the reviews is prepared, we can open a new chat with Claude and upload the file to the conversation.
After that, we can finally focus on analyzing the sentiment of the reviews about our business by feeding Claude the right prompts.
The following are some suitable prompts for this purpose. You can copy and paste them into your chat and modify them according to your needs.
Analyze the customer reviews in this document and identify the main emotions associated with the experience with our company. Identify recurring keywords and topics. Distinguish positive and negative polarities. Classify the reviews based on the prevailing emotion (e.g., joy, relaxation, disappointment, etc.). Measure the strength and frequency of these emotions.
Classify the sentiment of each review in this dataset as positive, negative, or neutral.
Generate a report showing the overall distribution of positive, negative, and neutral reviews.
What are the most frequently associated phrases and aspects with negative reviews?
Create a pie chart showing the prevalence of positive and negative sentiment over time.
Extract the key themes and topics most frequently discussed in positive and negative reviews.
What insights and suggestions would you propose based on this sentiment analysis to improve customer satisfaction?
Generate an executive summary report with the highlights of the sentiment analysis and your strategic recommendations.
Expected Results
The following are some of the responses obtained using these prompts. The reviews are from a restaurant.
Executive Summary – Restaurant Review Sentiment Analysis
The sentiment analysis of the 127 reviews collected over the past 6 years for the restaurant shows an overall positive sentiment.
Highlights:
82% of the reviews express positive sentiment, appreciating the quality of the food, the care in preparing the dishes, and the atmosphere of the venue.
12% of the reviews have a neutral sentiment.
Only 6% express negative sentiment, mainly criticizing the slow service.
The most frequent positive keywords are: excellent, good, quality, taste, refined, curated.
The most frequent negative keywords are: slow, waiting, disappointed.
The peak of positive reviews occurred 2 years ago.
Strategic Recommendations:
Maintain the high level of quality in the cuisine, a recognized strength from the reviews.
Monitor any service issues to promptly address them.
Leverage the largely positive sentiment for marketing activities, promoting the strengths across web channels.
Engage satisfied customers to generate positive word-of-mouth and reviews.
Systematically collect reviews to monitor sentiment trends and any emerging issues.
Analyzing the 8 negative reviews, the main critical aspects that emerged are:
Slow service (mentioned in 3 reviews)
Example phrases: “Slow even in getting the bill! We waited more than 50 minutes for a cheese and duck breast tasting”
“Everything very good, excessive waiting times.”
Disappointment compared to expectations (2 reviews)
“Terrible venue, poor service, chef’s menu that cooked what??? I wonder???”
“Very trendy place but not worth it, appetizers that left something to be desired, risotto not well-mantecated, and angus steak in the norm”
Small portions (1 review)
“Good dishes, the portions leave something to be desired. The owner is very generous. P.S. Watch out for the bill”
In summary, slow service and disappointment compared to expectations are the most recurring criticisms in negative reviews.
Here is the analysis of emotions and topics that emerged from the reviews:
Positive Emotions:
Satisfaction (52 reviews): keywords like “excellent”, “perfect” express a strong overall satisfaction.
Joy (32 reviews): words like “adore”, “sublime”, “excellent” denote a reaction of joy and enthusiasm.
Relaxation (12 reviews): words like “relaxing”, “quiet”, “intimate” reveal feelings of relaxation.
Positive Topics:
Food quality (67 mentions)
Staff professionalism (41 mentions)
Pleasant atmosphere (29 mentions)
Negative Emotions:
Disappointment (5 reviews): phrases like “disappointed”, “poor” show disappointment compared to expectations.
Frustration (3 reviews): words like “terrible”, “inadequate” denote frustration.
Negative Topics:
Slow service (4 mentions)
High prices (2 mentions)
In summary, the positive emotions of satisfaction, joy, and relaxation are clearly prevalent and associated mainly with food quality, staff professionalism, and pleasant atmosphere.
The negative emotions of disappointment and frustration, less frequent, are related to slow service and high prices.
This analysis provides useful insights to identify the strengths to emphasize and the areas for improvement to address in order to increase customer satisfaction.
Final Considerations
Automated sentiment analysis through artificial intelligence can be a fantastic help to quickly extract valuable insights from online reviews. However, it’s important to remember that the results should always be carefully analyzed and validated.
AI technologies such as automated scraping and natural language processing are continuously evolving and still have a certain degree of inaccuracy. Therefore, an expert human eye is necessary to critically interpret the results and place them in the proper business context.
In particular, it’s advisable to pay close attention to numerical and statistical references (percentages, quantities, etc.) present in the results generated by AI. These quantitative data often have limited accuracy, as current artificial intelligence models are not designed for precise mathematical processing.
These tools provide useful indications for identifying recurring themes and sentiments, but they do not completely replace human judgment. Companies must develop a strategy that wisely integrates automation with qualitative analysis by competent analysts.
In this way, the benefits of AI can be fully exploited to obtain valuable strategic insights, leading to concrete decisions and actions to effectively improve the customer experience.