How to use ChatGPT to summarize customer reviews?

How to use ChatGPT to summarize customer reviews?
Nowadays, it has become natural for many Internet users to leave a review about the products and services they use. By processing these customer reviews and delivering them to ChatGPT, you can Easily extract list of positive and negative points cited about any product or service, which you may find very useful for writing articles or other commercial and marketing uses.

In this article I offer you a very simple method to process customer reviews with ChatGPT Plus and ideas for using this data.

What can customer feedback collection be used for?

You can now use these reviews to create content with a very personalized dimension. Since Google’s Useful Content Update, an update aimed at promoting useful content, and more specifically those from 2023, we have observed that search engines place great importance on personal experience, dragging elements that demonstrate that we have an experience real of a topic. unlike an article that simply compiles information found on other sites. Using customer reviews allows you to highlight the strengths and weaknesses of a product or service.

You can also use customer reviews to Respond to customer or potential customer concerns if you have a business. For example, create content about problems or misunderstandings expressed by your customers.

Competitor Customer Reviews Can Help You guide a marketing and sales speech : Your weaknesses can be useful arguments to exploit in your own speech. Imagine, for example, that many of a competitor’s customers complain about the ergonomics of your products, and then you can strategically highlight the ergonomics of your own product so that it has more impact. In the same way you can enhance the strengths attributed to you.

These reviews also allow you better understand consumer languagean advantage to be more effective in SEO… or evenidentify ideas for new products, services or features.

Analyze customer reviews with ChatGPT

Where to find customer reviews?

For review analysis to be fruitful, it is best to choose platforms that include many reviews (at least a few dozen) and where ideally it is possible Sort them from newest to oldest. In fact, on many topics, recent opinions are the most relevant: a product or service has surely evolved over time and only recent opinions are a priori truly consistent with this evolution.

The idea is to recover the text of these reviews and export it to Excel format for example, so that the file can later be analyzed by ChatGPT. Any scraping extension data can do the trick, I use it for my part The Free Instant Data Scraper Chrome Extensionwhich has the advantage of being very ergonomic and easy to use, while allowing data to be extracted in Excel or .csv format.

Collect customer reviews on Amazon

Take Amazon for example. Let’s imagine that I want to write an article about a cheap bluetooth car kit. On the product page, navigate to the detailed list of reviews, ie. this pageand sort comments from newest to oldest using the available menu.

Sort Amazon Reviews
Sort Amazon Reviews

Then click Instant Data Scraper extension icon in Google Chrome.

Icône Instant Data Scraper

By default, it will highlight in yellow a part of the page where it thinks it can extract data.

Area highlighted in yellow

The idea is to click on the “Try another table” button until only the comments area is highlighted yellow, like this:

Selection of scraper area
Select the area to scrape

We see from the overview of the collected data that the extension achieves isolate review text in a dedicated column.

Données avis scrapées clients

Then click “Locate the Next button” in the extension pop-up window and click the “Next” button at the bottom of the Amazon comments area. This tells Instant Data Scraper how to get to the next comments pageso you can extract multiple pages at once.

next button

You can then increase the “maximum delay” (delay after which the extension stops pulling data), for example, to 60 seconds instead of the default 20 seconds.

Then click “Start crawling”, Instant Data Scraper will collect reviews on multiple pages. You can then click “XLSX” or “CSV” to download the data in the format that suits you best. Here, for example, I have collected about a hundred opinions over the last 4 months:

Collect Amazon Customer Reviews
Collect Amazon Customer Reviews

This is the file that we are going to ask ChatGPT to explode in a second time.

Collect customer reviews on Google Business Profile

Another example of data collection: Let’s say you want collect customer reviews this time on Google Business Profile (formerly Google My Business) to analyze reviews of a business. Type the brand name or “brand name + reviews” into Google to display your profile and click the number of reviews.

Google Business Review

Here you can also sort reviews from newest to oldest.

Sort the most recent reviews

Following the same principle as on Amazon, use the “Try another table” button if the highlighted area by default does not correspond to the reviews area… and mark the arrow that allows youshow more reviews by clicking on “Locate the Next button”. Increase the data collection timeout to 60 seconds or longer.

Show more Google reviews

Sometimes it is necessary to press the “Start Crawl” button several times in a row to collect as many reviews as possible; The table preview is enriched throughout the crawls with the new customer opinions collected.

Collection of customer reviews.

Export the file when you are satisfied with the result.

Now, let’s ChatGPT process this data… because obviously, it would be long and tedious to read each review by hand to extract general trends. It is in this context where artificial intelligence is very useful to save time!

What message should ChatGPT process these customer reviews?

Gonna ChatGPT and start a new chat, verifying that the most recent ChatGPT model is selected.

Upload your patch file and write a message like this:

Here is a file containing customer reviews collected on Amazon about a car bluetooth transmitter, the product with the title “Mohard Bluetooth Car, Bluetooth 5.3 FM Transmitter Wireless MP3 Music Player Radio Adapter, Hands-Free Calling, Dual 5V USB Ports /2.4A & 1A, car charger supports TF card/USB key”. Column I of the file contains the text of the most recent customer reviews. Can you analyze these reviews and highlight in list form the positive and negative points cited by customers about the product?

Upload a file to ChatGPT
Upload a file to ChatGPT

After some analysis of the file and some initial work of extracting and understanding some opinions, ChatGPT moves on to collecting your “comments” and the results are quite impressive, both in the positive and negative points.

Analysis of the positive points.

Analysis of negative points.

Pour analyze google reviewsThis time I use this message:

“Here is a file containing customer reviews published on Google for the company Blissim, which offers a beauty box and an e-store for beauty products. Column L of the file contains the text of the most recent customer reviews. “Can you analyze the reviews and highlight in the form of a detailed list the positive and negative points cited by customers about the company?”

Once again, the result is surprising:

Analysis of opinions about the Blissim company.

It is useful forprovide some context as I do about the company’s identity, so the analysis is more relevant if certain comments can be interpreted in several ways. Likewise, we could completely enrich the message by explaining why we intend this analysis or ask the AI to enrich its analysis with some quotes extracted from the analyzed opinions. You can also ask questions about what ChatGPT wrote.

I also advise you not to blindly trust the AI and that Please re-read the work provided by ChatGPT carefully. to check that you don’t miss interesting insights about the data.

I hope you find this method useful to analyze customer reviews with ChatGPT and use them in your articles or strategy.

Source link

Related Posts

Leave a Reply

Open chat
Scan the code
Hello 👋
Can we help you?