Keyword Analysis Using Excel

Data analysis with Excel is one of the most important developments in the information revolution. By using spreadsheets, we can manage, handle, and analyze vast amounts of data in a fraction of a second, transforming how businesses operate and make decisions.

But, processing data with Excel is generally limited to crunching numbers using functions, while analyzing text in spreadsheets continues to be a time-consuming task handled by humans.

With the advent of machine learning, text analysis models are now able to learn how to decode human language and analyze huge batches of text content to get new insights from data.

Keyword analysis is a great example of AI-powered data analysis. With a keyword analysis model, customer support, UX or product teams can quickly sift through spreadsheets filled with text from survey responses, customer feedback, social media content, and other text data, and automatically detect the most relevant keywords from the content. A keyword extractor can save countless hours of manual work and deliver results that will help you get insights about what words or expressions are being associated with your products or services.

The best part is that this data revolution has reached a point where there’s no need to be a programming or AI expert to enjoy the benefits of machine learning for analyzing text. You can simply upload your Excel spreadsheets to a keyword analysis model today and let it do the heavy lifting for you.

In this post, we’ll cover what a keyword extraction model does and how you can use MonkeyLearn to do keyword extraction in Excel. We’ll share some handy tutorials, as well as some popular uses and applications for this technology:

Let’s get started!

What is Keyword Extraction?

Keyword extraction (AKA keyword analysis or keyword detection) is a text analysis technique used to understand the most relevant keywords from text data automatically. Keywords can be compounded by one or more words, and are defined as the important topics mentioned in the content.

For example, take the following text:

Elon Musk has shared a photo of the spacesuit designed by SpaceX. This is the second image shared of the new design and the first to feature the spacesuit’s full-body look.

The most relevant keywords from this text might be Elon Musk, Photo, Spacesuit, SpaceX, Second Image, New Design, and Body Look.

Extracted keywords can be used for tagging rows of data with the most relevant terms and creating reports and visualizations for discovering valuable insights.

Keyword Analysis in Your Excel Spreadsheets

Let’s go over how to use a pre-trained keyword extractor with MonkeyLearn, a text analysis platform that allows users to use a text analysis model in a few simple steps:

This pre-trained extractor will enable you to analyze your spreadsheets for keywords without the need to train your own model, providing a ready-to-go experience.

Step 1: Go to the keyword extraction model

Sign up for free to MonkeyLearn and go to the ‘Explore’ tab, filter by ‘Extraction’ and click on ‘Keyword Extractor’:

2. Upload your Excel spreadsheet

Next, upload the spreadsheet you want to analyze and perform keyword extraction by following this path: Run > Batch > + New Batch:

Once you’ve selected and uploaded your Excel file, you’ll need to select the column or columns you want to use for the keyword analysis:

And you’re done! After MonkeyLearn has processed the data you’ve uploaded, it will download a spreadsheet to your system with an extra column for the keyword extraction results:


Get More Accuracy with a Custom Extractor

This pre-trained keywords extractor is a great way to get you started right away with automatically detecting keywords with machine learning. But, keywords are subjective, A particular word within a text might be or might not be a keyword depending on the context and use case. In this case, having a custom extractor trained to detect keywords and phrases adjusted to your particular business needs might prove to be much more effective.

For instance, let’s use this HubSpot customer feedback piece as an example:

“In short, AMAZING! Not only was the UI easy to learn, we were able to sync our outbound emails to automatically create new contacts based on the recipients’ domain. Having said that, the pricing is quite expensive and it's a blocker for us. We are still a small startup and we don't have the budget to pay the full price once our discount is over."

These few lines of text contain several different keywords that may be useful for different teams. Let’s say we are part of HubSpot’s product department, then keywords like UI, Easy to Learn, Sync and Emails will probably be the ones we’re looking for.

But, if we’re in the sales department, these won’t be of much use to us. Our keyword extraction model will probably be oriented towards detecting things like Startup, Budget, Full Price, and Discount.

This means that a custom extractor can be trained to highlight areas of interest for each team. If you feel like customizing your own keyword extraction model, then you can always do it through MonkeyLearn. Here’s a handy tutorial on how to do exactly that.

Use Cases and Applications

Keyword analysis in Excel can be used in a vast number of businesses. From analyzing customer feedback to social media conversations and support tickets, any piece of text containing valuable insights can be analyzed in no time, without the need for time-consuming manual input. Let’s take a look at how keyword analysis can deliver insights to your various teams:

  • Product Feedback

Net Promoter Score (NPS), customer surveys, comment sections, and social media entries can be analyzed for identifying relevant keywords at lightning speed to get the insights you are looking for.

Customer feedback can come in many shapes and forms, and every single one of those contains vital information that can help your team prioritize tasks. With keyword extraction, you can get an overview of what your customers are talking about in relation to your products or services.

For instance, imagine HubSpot analyzing their last 1,000 product reviews, only to discover that the most common keywords are Bad, UI, Bugs, Slow, and Confused. This would be a strong indication that the user experience isn’t up to scratch or that the interface needs to be worked on to make it easier for customers to navigate the platform. Slow and Bugs would most likely be related to the overall performance of their product, something they would get their dev team to check out as soon as possible.

Your dev team’s roadmap can be planned by checking out the results from keyword extraction applied to those cramped Excel spreadsheets. There’s a cool report published by that shows how they visualize the results of a keyword extractor with NPS responses:


In this graph, we can easily see the results of using keyword extraction to understand trends and common topics in customer feedback for This batch of around 1,000 NPS responses not only shows that Service and Quality are the two most used keywords when expressing their opinion. By cross-referencing their Net Promoter Scores with a keyword extraction model, they were able to link promoters, passives, and detractors with relevant keywords. This graph shows that Price is a keyword that carried the most passives and detractors, together with Phone. Now, knows that they need to work on these aspects of their service in the near future.

Furthermore, the guys at can be proud about their work with the overall service they provide, since keywords like Convenience and Convenient Service are only mentioned by their promoters, visualized in this chart as fully green bars. Excellent Service and Speed also get great feedback from their customers, further solidifying the notion that Promoter’s main product and its support staff are doing a fantastic job.

Now, something’s team should clearly work on is how their surveys look on mobile, which shows different satisfaction rates stemming from their service. Phone holds an alarming percentage of detractors, probably linked to the mobile responsive design of Promoter’s NPS surveys.

You can also check this neat step-by-step example of how a SaaS team can use text analysis models to analyze product reviews and get useful, actionable insights. Below, you’ll find a chart showing several categories related to Slack’s service, the number of times customers mentioned each category and the overall sentiment applied to them as a result of a sentiment analysis check:


From this chart, it’s clear that Slack is doing pretty well in the vast majority of their service features. Some alarms should be raised in categories like Messages and Performance-Quality-Reliability, where the prevailing sentiment is not only Bad but is present in almost 1,000 reviews.

Further analysis using a keyword extractor on negative reviews showed which keywords were used the most by their customers when criticizing their service. The key phrase Paid Version had an alarming relevance when analyzing the data, showing that their premium service might need some tweaking to meet their user’s expectations. App, Messages, and Notifications were also present at the top of the negative list, perhaps indicating that Slack’s mobile app needs to be updated and that the problem might be in the messages and notifications area.

These two examples show how your team can use keyword extraction to get the most out of your customer’s feedback by learning how often they mention specific aspects of your service, but also using sentiment analysis to get more fine-grained insights that will help you plan and effectively direct your efforts and resources.

  • Market Research

With a keyword extractor, you can also monitor any text feed for a day-to-day understanding of what’s hot and what’s tanking in your industry.

Is a particular brand engaging in a new way with your target audience? Are your competitors ahead in awareness, or have they just launched a new product or service?

Your team can take note of all this information by uploading an Excel file with text data and automatically analyzing it with a keyword extraction model.

For instance, check out this post in which we compiled over 1 million hotel reviews around the world and tagged them by different aspects like location, stars and fine-grained criteria that not only divides positive and negative feedback but also corresponds to the aspect provided by each hotel.

Fascinating conclusions stem from the analysis. For instance, after narrowing our search for just negative reviews on _Cleanliness _relating to hotels in New York, and running a keyword analysis, we were able to determine that most of those reviews referred to the following keywords:

  • room
  • bathroom
  • carpet
  • towels
  • bed bugs
  • bed
  • hotel
  • shower
  • shared bathroom
  • walls

A customer satisfaction team working for a hotel in the New York area would now be able to stand out from their competitors by providing exactly what other hotels are missing, and critically failing to deliver.

Another interesting insight obtained through this extensive analysis was the fact that keywords relating to cockroaches were only present in reviews for hotels in Bangkok.

  • Brand Monitoring

By narrowing your search to just your own company or specific competitors, your team can get useful insights about your brand’s perception within your industry. Add relevant specs as keywords to your extractor and learn which aspects of your service needs to be revamped or completely redesigned.

For instance, executing a brand monitoring strategy through the vast mass of content provided by Twitter can be done automatically and result in useful insights for any company. Take this example, in which we did a feedback comparison of mobile carrier companies like AT&T, Verizon, Spring, and T-Mobile.

Apart from the model employed to identify the most relevant keywords, a sentiment analysis model was also put to the test to understand how positively, negatively or neutrally these customers were talking about these brands.

Interesting insights were identified when analyzing the results of this procedure. One of them was related to T-Mobile’s customer support team. Tweets positively mentioning this particular telco often contained keywords that were names of members of its support team. This means that their customers were truly engaged by their experience, actually remembering and praising the work of individuals inside the company.

In comparison, tweets referring to Verizon mentioned common keywords such as Better network, New phone, Rewards, Thanks, and Quality Customer Service, without any topic standing out or indicating an engaging experience.

Final Words

Keyword extraction can completely change your Excel experience. There’s no need to spend your budget on hiring developers or AI experts in this day and age. With tools like MonkeyLearn, you can use a pre-trained extractor to get started right away to detect keywords from spreadsheets, gain more insights from your data, and enhance decision-making.

If you are looking for something more personalized, you can build your own keyword extractor in a few minutes and start increasing your team’s productivity by erasing time-consuming tasks.

Want to see keyword analysis models in action? Request a demo to learn how your company can make the most of machine learning tools.

Federico Pascual

April 23rd, 2019