How to Use AI in Excel for Automated Text Analysis

‘Insight’ is a word we hear too often when we’re running data analysis. Whether we are trying to understand the voice of the customer or monitor our social media channels, it’s insights that we’re trying to gain to make better decisions.

The thing is we try our best to understand our audience by crunching the numbers, but sometimes that’s not enough to find the insights we are looking for. 

Heaps of text is created every day in the form of social media comments, emails, chat conversations, customer surveys, product reviews, and the like. And all this data contains valuable insights. As new batches of text appear, we diligently read and analyze these texts, constantly wondering if there’s another way to gain those valuable insights without sifting through huge amounts of text and repeating the same processes over and over again. 

By using text analysis powered by AI, you can start gaining insights without monotonous and time-consuming manual processes. This technique based on machine learning can automatically analyze text in your spreadsheets, so you don’t have to. Tools such as sentiment analysis, topic detection, keyword extraction can get the information you need to make data-driven decisions.

If you have survey responses, product reviews, or social media mentions on Excel, you can receive the text analysis results automatically in that same spreadsheet. Don’t worry, it’s not complicated, and here you’ll find a step-by-step guide on how to use text analysis within Excel and how to get the answers to questions like:

  • What do customers like and dislike about your product?
  • How many positive mentions do you get on social media?
  • What issues do you need to address right away?
  • What words or expressions do customers use to describe your brand?

Text analysis helps you the bigger picture of your business, but first, you’ll need to lay the foundations. In this guide, we’ll dig into the following topics so you can get started with text analysis in Excel:

Let’s get the ball rolling.

What is Text Analysis?

Text analysis, also called text mining or textual analysis, is the automated process of classifying and extracting information from text using AI, whether it comes from emails, tweets, blog posts, or product reviews. This means that a text analysis model can read text, for example on an Excel spreadsheet, and structure it automatically.

AI and machine learning may not sound like a familiar concept, but it’s something we come across more often than we realize. Just think of Gmail response suggestions. How does the app suggest responses, such as “Thanks”, “I’ll check it out” or “No problem” to an email your colleague sent you with? That’s text analysis detecting words and expressions within emails, categorizing them and suggesting appropriate responses. There’s no employee behind the curtain typing suggestions manually. 

Now, can text analysis be used for other applications? Yes, this technique can be applied for different use cases, such as identifying the sentiment behind a survey response, the urgency of a customer ticket or the most used expressions in social media mentions.

For these and other purposes, there are different techniques close at hand. We’ll cover them in a bit!

How Can Text Analysis Be Useful?

Think about how many emails, chats, tweets, survey responses, product reviews, and support tickets you receive on a daily basis… to say that a big part of your business data is text-based, is an understatement.

Gathering data in an Excel spreadsheet, analyzing it and attempting to obtain insights from text is a standard process within most businesses. But with the increasing influxes of data and the time required to manually analyze text, businesses often fail to keep up with new data and, as a result, fail spot insights and trends that could take them to the next level.

By incorporating text analysis with machine learning, detecting trends and insights from text data is easier than ever:

  • Scalability

    You read row by row trying to obtain insights from each text but there’s just too much information to read. How many hours would you need to sort through all the data you gather on your Excel spreadsheet on a daily basis? Definitely too many. With text analysis, hours can be scaled down to just a few seconds so you can make better use of your time. 
  • Real-time analysis

    You’ve just gathered the latest social media comments on your spreadsheet. You notice a surprising number of negative comments. What’s the problem? It turns out you failed to spot some mentions about a technical issue with your app. The complaints piled up and, as a result, your customer experience suffered. Plus, your slow response time even led to customer churn. Could this have been prevented? Text analysis can be used to detect critical issues automatically in social media. This way, you have the information to act right away when keywords such as Bug or Problem start to rack up. 
  • Consistent criteria

    Let’s say you’ve gathered a batch of customer feedback from social media on your spreadsheet, and your team needs to classify them as Positive, Neutral, or Negative. Our ability to discern differences between text can fluctuate from one day to the next, we might feel tired and hesitate, or simply make mistakes and even change our minds. Different views are not going to deliver accurate and insightful results. So how can we make sure that our analysis criteria remain consistent? Text analysis uses just one set of criteria – the criteria you use to train a model – which it applies consistently. No yawning, no hesitation.

Now that you know the main benefits of using text analysis in your Excel spreadsheets, you may be wondering what kinds of analysis you can run. This will mainly depend on the insights you hope to reveal. 

Let’s go through your text analysis options: 

Text Classification

Text classification is the process of tagging pieces of text by category depending on their content. Text classification covers a wide range of applications, for example, you can identify the topic of a given text, its sentiment or intent, and even its language

Topic Analysis

Topic analysis assigns tags based on themes or topics. This basically means that if you receive a survey response that says ‘Clean and easy-to-use user interface, I love it!’ , it would probably be tagged as UX, whereas a survey response that reads ‘Your team has been super helpful!’ would be categorized as Customer Support.

How can this be useful? Imagine that you want to reveal the most mentioned topics in the feedback you receive from customers (e.g. in customer surveys or public reviews). By running topic classification, you will get an idea of what things customers talk about when referring to your product. Is there a recent spike of customer feedback related to UX? Why? By looking into them, you may learn what you are doing right and what needs to be improved in your user experience. 

Additionally, customer feedback tagged under UX could be automatically routed to the dev team, and the same could be done for other tags: feedback tagged as Billing would be sent to the team in charge of finances, and feedback under Plans & Pricing would be sent to the sales team.

To get a glance at how a topic classifier works, you can check out this pre-trained NPS SaaS Feedback Classifier, below, which tags NPS responses into categories such as Customer Support, Ease of Use, Features, and Pricing:

Sentiment Analysis 

Sentiment analysis is the automated process of assigning tags according to how people feel about a certain subject (e.g. positively or negatively).

What would you normally do if you want to know how people feel about your product or service? Maybe manually go through each survey response or social media mention and classify them as Positive, Neutral, or Negative. That would work, but it’s not very efficient. Can you imagine gathering and tagging hundreds of mentions every week? Sentiment analysis is a technique that does exactly that, but faster and automatically. 

You can try out a pre-trained model here which classifies text as Positive, Negative, and Neutral:

By combining sentiment analysis with topic classification – something called aspect-based sentiment analysis, you’ll not only be able to understand what a text is talking about (topics) but also how (sentiments). This can be useful for getting a more complete picture of the data in your spreadsheets. For example, by using aspect-based sentiment analysis on customer feedback you can understand if customers are praising the UX, but complaining about your customer service.

Intent Detection

What does the customer want? That’s a question businesses crave to answer and it’s not an easy task. Of course, machine learning techniques can’t give you the exact answer, but they can point you in the right direction of what your customers need.

For instance, you receive emails from potential customers that can automatically be classified as Interested or Not Interested. Then, you can send a follow-up email only to those classified as Interested. That’s the case of this pre-trained model that also includes other intents such as Subscribe, Unsubscribe and Email Bounce:

Language Detection

If you have a big team and customers that speak different languages, you may face some difficulties finding the right team member to read customer surveys. Maybe you receive a dozen responses and start reading them, only to find out that they all need to be assigned to a team member that speaks the language they’re written in. Ok, it may not sound like a great inconvenience, but it can be an unnecessary step in the process.

Language detection can identify different languages and assign tags for each of them so that the next time you receive a customer survey response, you’ll be certain that it’s one you can understand. As far as the other responses go, they can be automatically routed to the team that speaks the language to speed up the process. 

If you want to take a look at language detection in action, here’s a pre-trained model that can recognize 49 different languages:

Text Extraction

Text extraction is used to extract data that’s within a text: keywords, prices, company names, entities, locations, etc. While a text classifier helps sort text into different categories or ‘buckets’ according to their content, a text extractor can obtain specific pieces of data that exist within the text.

Let’s take a look at the main text extraction models to get the idea:

Keyword Extraction

A keyword extractor can be used to obtain the most important words or expressions from a piece of text. It works by identifying the words and expressions that are the most representative within a given text, and delivers them as values, or results, which act as a summary of the text.

For instance, if you have a batch of survey responses or product reviews on your Excel spreadsheet and you’re trying to pinpoint how many times a product feature that you just launched is mentioned, you can run a quick analysis and get the numbers within seconds. 

Here’s a pre-trained keyword extractor you can take a look at. Just type your text and see how it goes:

Entity Extraction

An entity extractor obtains names of people, companies, brands, and more from a given text. 

Let’s say you want to run an analysis on a product review website to see how often your competition and brand are mentioned. You can gather your information in an Excel spreadsheet and then run an analysis to get the results. 

Also, if you have many branches in your country or worldwide, you could use an entity extractor to identify complaints. So, if someone tweets “The Florida branch doesn’t have enough staff”, after running the extraction and finding that “Florida branch” is the entity mentioned, you can send the complaint to them. 

Here’s a pre-trained entity extractor you can try out: 

How to Use AI Text Analysis in Excel

Ok, now that you’ve got the gist of text analysis, how can you start applying it to the data you have on Excel? Will it take a long time?

If you want to get insights from the data in your spreadsheets ASAP, you will only need to follow a few simple steps with the pre-trained text analysis models that come with MonkeyLearn. These models have already been trained, so the only thing left for you to do is get the analysis.

Here’s how:

1- Choose a Model

Once you’ve signed up for free to MonkeyLearn, it’s time to explore your options. Just go to your dashboard and click on explore. Different pre-trained models for text analysis will come up, including models for sentiment analysis, keyword extraction, topic analysis, language detection, and more: 

Now it’s time to pick the right model for your needs. For this tutorial, we’re going to go through the step-by-step process of using a pre-trained sentiment analysis model, but the process is identical for all of them.

Go ahead and filter the available models by Sentiment Analysis and then click on the first model:

2- Upload your Excel Spreadsheet

Once you have chosen your model, go to the ‘batch’ section and click on ‘new batch’ then ‘browse file’ and upload the Excel spreadsheet with the information you want to automatically analyze with the model:

Keep in mind that MonkeyLearn also accepts CSV files and, alternatively, you can upload training data from third-party tools such as Zendesk, Gmail, and

3- Take a Look at the Results!

A few seconds after uploading your Excel spreadsheet, a similar file will automatically download to your computer. Open it up and you’ll see the predictions next to each row of text:

Voilà! Now you can use text analysis models to automatically analyze text in Excel.

Create Your Own Machine Learning Model

Yes, you’re ready to start getting those insights from text analysis! The thing is, what do you do if you need a model that can predict results specific to your business? 

This may be necessary if you want to categorize customer feedback using specific categories, if you want to use your own criteria for sentiment analysis, or if your industry has unique keywords or expressions that aren’t detected by a pre-trained keyword extractor. It’s in these instances that creating a custom model may be your best option. 

Basically, you’ll have to train a model with your data so that it can learn from your criteria. It’s pretty straightforward and can provide more accurate results than using a pre-trained model.

Take a look at the following steps for creating a custom classifier with MonkeyLearn:

1- Create your Model

This time we will create a model from the ground up, so select ‘create model’ in your dashboard:

Now it’s time to choose the type of model you want to build. In this case, we’ll go for a classifier:

2- Upload your Data

We need to upload an Excel spreadsheet with the text data that we’re going to use to train the model. Besides Excel, you can upload data using a CSV file or third-party tools such as, Front, or Gmail:

3- Create the Tags

After uploading the training data, define the categories you want to use in your classifier: 

Take into account that the more tags you have, the more training data you’ll need. 

4- Train your Model

It’s time to start teaching your model how to classify text according to your criteria. This may take some time, but it’ll be worth it. Just select the correct tag for each text example and click on ‘confirm’:

You’ll notice that your model starts making predictions on its own that you can either confirm or correct. Continue tagging until you all predictions are spot-on. 

5- Test your Model

Is it ready for production? You can find out by testing the model. After doing some tagging, MonkeyLearn will ask you to name your model and give you the option to ‘Keep Training It’ or ‘Test It’. Let’s select the latter option: 

Write something to see if your model classifies it correctly: 

Not getting accurate predictions? Don’t worry. It may just need some more training. Just go back to the ‘build’ section and tag more data. After a while, click on the ‘run’ tab again to see how it’s doing.

6- Put the Model to Work

The last step is using your trained model to analyze new data. To do this, just upload your Excel spreadsheet by clicking on ‘new batch’. Select the file and in a few seconds, you’ll receive a similar document with a column of predictions!

You’re done! Now you have a custom classifier that can classify text for you!

Training a custom extractor involves a similar process. Follow this tutorial for a handy step-by-step guide to getting started with custom extraction.

Use Cases & Applications

Now you can analyze text at scale in Excel without endless hours of hard work, but what now? There are many ways in which text analysis can serve your business, depending on what you want to achieve. Let’s take a look at some ways businesses are already using this handy tool.

Get Insights from Customer Feedback

You ask your clients for feedback and you start to receive survey responses… but more than you expected. Soon they start to pile up in your spreadsheets and you are having a hard time trying to make sense of it. However, with text analysis you can whip through all these responses in no time at all.

Running a sentiment analysis can shed some light on how many satisfied customers you have, and if you combine it with a topic classification not only will you learn how customers feel about your brand, you’ll also discover what they’re talking about.

Retently used topic classification to automatically analyze open-ended responses and categorize the feedback into categories such as Product Features, Product UX, Customer Support, Integrations, Pricing, Ease of Use, and others:

As a result of running the topic analysis, Retently noticed that happy customers often talk about Customer Support, but unhappy customers often mention Product Features.

In contrast, used keyword analysis to learn which words are linked to positive and negative feedback:

Understand Your Brand Perception 

What do people think about a brand? We can dig into social media comments, new articles, and product reviews to learn how people feel about a specific product or service. 

As an example, we used text analysis to better understand the Twitter conversation around the US Presidential election from 2016. We used sentiment analysis to understand how both candidates were perceived. The following are the Twitter mentions for Donald Trump, classified by sentiment over time:

As a contrast, these are the Twitter mentions for Hillary Clinton: 

These charts show the peaks, the valleys and the troughs of the candidate’s perception and it can do the same for your brand: just use the sentiment analysis model to analyze your social media comments over time. You’ll be able to spot when the positive comments increase or decrease and understand your brand perception over time.

Want to understand what your customers are saying, and not just how they feel about your brand? Then do what we did with Slack and run an aspect-based sentiment analysis to get something like this: 

By looking at the results, you can easily detect that people have a problem with how notifications in Slack work. How can this be resolved? Well, that’s for the Slack team to decide, but it’s definitely easier to monitor how customers feel about different aspects of their brand, and as a result, they can take action if they consider it to be important.

Now that you know how to gain insights about what people like or dislike about your brand, you can start making data-driven decisions that take your brand to the next level.

Do Valuable Market Research

Staying at the top of your game to stay in the game; that’s one of the main rules businesses follow and that’s why researching trends is essential. Is there a new need you could be fulfilling? Are customers opting for a different service because yours doesn’t provide something?

If you run a text analysis on social media posts, public reviews, or news articles, you may find an unexplored niche and take the necessary steps to stay at the forefront. 

It may take you hours or even days to do this research and analysis manually, but with a keyword extractor, you can easily spot the main topics your target audience is talking about and maybe come across an unexpected growth opportunity.

For example, you could first find out how many positive mentions your company has on Twitter compared to your competitors’. Take a look at these results for positive mentions on Twitter for four different US telcosTweet mentions for four different US telcos

Now, take a look at the negative and neutral mentions: 

When we ran a keyword extractor to learn what words were used for positive and negative comments for each brand, we were able to find out which complaints are unique to a single carrier, for example, T-Mobile was the only carrier that received complaints about their LTE service. In contrast, Verizon received negative mentions about their ‘Unlimited Plan’, which apparently isn’t unlimited. 

How can you benefit from using text analysis for marketing research purposes? Maybe you find that positive comments regarding your closest competitor are mostly related to Functionality. Perhaps this would encourage you to improve that area to stay ahead of the game. Or maybe you notice that your competitors’ customers complain a lot about customer service. This might be an opportunity to use your awesome support to your advantage, and make it a central piece of your brand positioning.


Understanding what the customer is saying has been a challenge for a long time. We read information on Excel to see what we’re doing right and what we’re doing wrong and while we read, we lose time that could be better spent on taking action. 

Instead of spending hours manually analyzing rows of text in your Excel, now you can use text analysis within your spreadsheets to get the insights you need to make data-driven decisions in a cost-effective way.

If you want to learn more about how to use text analysis with Excel, you can request a demo. Our team will guide you on how to use text analysis in Excel and apply this technology in your business.

Federico Pascual

Federico Pascual

COO & Co-Founder @MonkeyLearn. Machine Learning. @500startups B14. @Galvanize SoMa. TEDxDurazno Speaker. Wannabe musician and traveler.


Have something to say?

Text Analysis with Machine Learning

Turn tweets, emails, documents, webpages and more into actionable data. Automate
business processes and save hours of manual data processing.