Last week I wrote about my first project with Keatext where I was looking for ways to enhance the user experience. This week I am putting together a tutorial for new users. The purpose is to give you an idea how to use Keatext on your first project.
Analyzing your data with Keatext
Step 0. Preparing your data
Keatext accepts data in csv or from external sources like Salesforce or Zendesk. For my project I chose to upload my data in a CSV format. Make sure every column is clearly labeled on the first line.
Step 1. Log in
Step 2. Set up your Project
You need to give your project a name and an optional description. This will be visible to the colleagues you choose to collaborate with so try to be as descriptive as possible. Next, I linked my data set. I made sure “Create new dataset” was selected. I named my data and choose the file on my computer to upload it.
Keatext automatically detects the type of data you have in each column. Do double check it, especially when it comes to dates and numbers. If the field was not detected, there may be some issues with your data.
Next up is an important step: ticking Analyze vs. Filter.
I have four fields: Name (the brand the review is about), Review Rating (a 1-5 rating), Review Title and Review Content.
I only want to tick “Analyze” on the fields containing text, so that’s “Review Title” and “Review Content”. I also want to be able to filter by “Name” and “Review Rating”. You then click the “Link Dataset” button. Depending on the size of your data you may need to be a little patient until Keatext analyzes it. Mine took about 3 minutes since I had 3,300 entries. That’s pretty magical considering it would take a human about 27 hours (30 seconds/review) just to read all the reviews, not to mention process them.
If you want to try this out for yourself, there’s a similar walkthrough on this section.
Step 3. Grouping
If you have a lot of data, you can already start grouping topics even if the analysis is not ready. Just be mindful that further variation can show up after the data is fully uploaded. Grouping allows you to tell Keatext that different terms synonymous for the purpose of your analysis.
You can find the group functionality by pressing on the third button in the upper right corner. This is highly recommended on a small dataset like mine, but generally it makes the topics tidy.
A great example of words that needed to be grouped is “money” since there are so many variations.
I only bothered grouping words that are above a certain frequency, no need to do too much work on words that are statistically insignificant. Note that groups can be exported to future projects if needed.
Step 4. Exploring the data
Here I share some ideas on exploring the data, but my list is by no means exhaustive. The idea is to get familiar with the information and start picking up on the possible stories it might be telling.
First, you can have a look at the most common terms. You can click on each one to see an overview of the topics that are related to it and the sentiment analysis related to this word.
If you need to dig even deeper you can press the “View records” button to get to the records of the mentions.
I used the records to clarify what the reviews meant by certain words, find clues as to what fuelled their sentiments and see how the relations with other terms were formed. It’s framing the qualitative analysis in a very clean way.
You can also sort topics by sentiment.
With this feature, I can see what topics were most prevalent in the negative or positive reviews. I found that in both cases they were talking about similar topics which was an interesting find in itself.
This is a filter created by your data. In my case I have two options. The first is to filter by review rating. This can be very powerful when combined with sentiment analysis since reviewers can still include negative feedback in high rated reviews or positive feedback in negatively rated reviews.
My second filtering option includes the name of the platform being reviewed. I imagine this could come in very handy for competitive analysis.
Step 5. Extracting insights
After exploring my data with our intuitive interface, I started noticing some stories the data was telling. I’ve gone through my thought pattern in my previous post.
To sum it up, noticed that two main topics survey and money were strongly related to words describing time.
After further investigation it was obvious that survey platforms needed to pay attention to the user experience when filling out surveys and in making sure the payment system works in a timely fashion.
Step 6. Reporting
You can add charts that highlight important insights.
You have a few options, including “time series”, which is perfect if your data contains time stamps, enabling you to view trends over time. You can export charts as pictures or send them to a board.
As I just mentioned, boards are great for collecting all the best graphs. You can make boards for different types of colleagues and share it with them by using a custom link. You can also export the data analyzed by Keatext in a CSV straight to your email. This can be useful if you’re using advanced visualization tools like Tableau.
That’s the end of my tutorial. I hope this article gave you a better idea of how to use Keatext for the first time. I’d love to hear back from you if this kind of tutorial was useful. Let me know in the comments below.
FRESH CONTENT DELIVERED STRAIGHT TO YOUR INBOX
- How sentiment analysis can improve your customer experience (CX)
- Scaling Corporate Culture Assessment with Artificial Intelligence
- Choosing the right Text Analytics partner for your VoC platform
- VoC platforms: Choosing the right Text Analytics technology
- Survey Platforms: The pitfalls of building text analytics in-house