Use Cases

Intent Classification: How to Identify What Customers Want

Intent Classification: How to Identify What Customers Want

Intent classification is the automated association of text to a specific purpose or goal. In essence, a classifier analyzes pieces of text and categorizes them into intents such as Purchase, Downgrade, Unsubscribe, and Demo Request. This is useful to understand the intentions behind customer queries, emails, chat conversations, social media comments, and more, to automate processes, and get insights from customer interactions.

For example, identifying purchasing intent is pivotal in transforming sales leads into fully-fledged customers. And the key to it all is timing. Leads move fast, and even more so in the world of online sales where everything seems to happen at lightning speed. It is critical, therefore, for businesses to address and respond to prospects quickly because it increases their chances of closing sales.

As stated in the Harvard Business Review, responding to potential customers within an hour of receiving a query increased the chances of a business having meaningful conversations with key decision-makers, by nearly up to seven times.

That’s something you don’t want to miss out on. 

However, manually detecting customer intent from user interactions via email, chat conversations, or social media posts is a slow, time-consuming, and unreliable process. 

But fear not. Automated intent detection with artificial intelligence can help identify intent in real-time, giving you the opportunity to reply to potential customers instantly.

In this post, we’ll cover the intricacies of intent classification – how it works and its importance – along with a tutorial on how to create your own intent classifier coupled with relevant use cases and applications.

Read along, or jump to the section of your interest:

Let’s get rolling.

What Is Intent Classification?

Inherently linked to Natural Language Processing (NLP), intent classification automatically finds purpose and goals in text. For example, imagine that you are interested in subscribing to cloud service pCloud and drop them a line asking about their paid subscription plans:

Hi, I’m a photographer and work with a significant amount of raw files. What kind of storage do you offer? Is it a lifetime membership? For the right price, I’d love to purchase cloud storage.

With an intent classifier, you could easily locate this query among the numerous user interactions you receive on a daily basis, and automatically categorize it as a clear Purchase intent.

Ultimately, the goal of intent classification is to help you pinpoint the exact motivation behind pieces of text. Every customer interaction has a purpose, an aim, or intention. Whether it’s purchase intent, a request for more information, or someone who wants to unsubscribe, you should be able to respond to sales leads quickly to increase your chances of closing the sale.

How Does Intent Classification Work?

Think about how humans classify everything – jeans are an item of clothing, a guitar is an instrument, a song with a soft rhythm is relaxing.

In the same way, an intent classifier is able to categorize text based on the intent, goal, or purpose expressed in its content. It does this by using machine learning algorithms that can associate words or expressions with a particular intent. For example, a machine learning model can learn that words such as buy or acquire are often associated with a Purchase intent.

However, these machine learning classifiers need to be trained first with examples. First, you need to define tags or categories that are relevant to the matter at hand. For intent classification, these tags typically refer to actions that customers intend to perform. 

For example, if you’re analyzing customer emails, your tags might be something like Interested, Need Information, Unsubscribe, Wrong Person, Email Bounce, Autoreply, etc.

With tags in place, you can begin to train your model and feed it relevant examples for each tag. This way, you’ll teach the classifier how to tag new data appropriately. For example: “I tried to make a purchase through the site but I don’t know where to start, could you help me out? I’m really interested in shopping the new collection.” This email can be tagged as Interested.

Now, imagine that you are doing outbound sales and reaching out to potential customers via email. Eventually, you may receive a response like “I’m not the right person for evaluating this solution, usually our manager gets to decide which tools we use.” This email can be tagged as Wrong Person.

You may also receive a response such as “I am out of the office until next week. I’ll try and reach you once I get back. Thanks.” This email can be tagged as Out Of Office.

The more examples you provide the model, the smarter your intent classifier will be since it has more information to learn from.

To take intent detection one step further, you can combine an intent classifier with the results of a text extractor. This type of text analysis model can identify specific data from text, such as locations, dates, organization or person names, that are related to a user’s intent.

For example, if you receive a message like, “I want to book a flight from New York to Las Vegas, but my card has been declined” an intent classifier would categorize this message as intent to Book a Flight, and a text extractor would extract the entities New York and Las Vegas.

Why Is It Useful?

Identifying a customer’s purchasing intent is pivotal in transforming sales leads into fully-fledged customers. 

Intent classification allows businesses to be more customer-centric, especially in areas such as customer support and sales. From responding to leads faster, to dealing with large amounts of queries and offering a personalized service, intent classification can be a key element in your business. Let’s take a closer look at some of the benefits:

Take Advantage of Every Sales Opportunity

Automatically detecting purchasing intents quickly, with the help of artificial intelligence, is crucial for sales and customer support because it allows companies to take immediate action and transform leads into paying customers – or prevent churn. The faster teams can detect purchase intents and respond, the more chances they have at closing a deal. 

Prospective clients appreciate fast-paced responses, with some expecting a response in less than 6 hours. Let’s say you receive a Facebook message asking for product availability. With an intent classifier, you can quickly identify this interaction as an interested client and contact them immediately to boost the possibility of turning this engagement into a sale.

Scalability

Even when companies are inundated with data, intent classifiers are able to pinpoint potential customers who have expressed interest and direct these specific queries to the sales teams. Machines work faster than humans, non-stop, and they don’t grow tired – so even as workloads increase, they’ll never miss a potential sale! 

Consistent Criteria

Machines are consistent in the way they operate, processing data using the same parameters and criteria for every single instance. The consistency in criteria ensures that all the customer intents are analyzed under the same circumstances, applying the same standards, protocols, algorithms, etc. This greatly reduces errors and improves data accuracy.

Increase Conversion Rates in Sales Campaigns

As you deploy a marketing campaign and start receiving customer interactions, you can use intent classifiers to identify potential customers that show a ‘high intent’ of purchasing and contact them immediately. Thus, your conversion rates skyrocket.

Get Analytics From Sales Campaigns

With clear intents identified automatically in your sales and marketing campaigns, you could easily create reports based on factual data about conversion rates, interested buyers, upsell opportunities, and much more.

How to Get Started with Intent Classification

When you bring machine learning into the conversation of intent classification, some people shy away from it as they believe there’s a lot of programming, mathematics, and other complex sciences involved. And to an extent, they’re right. 

Fortunately, there are ready-to-use tools, such as MonkeyLearn, that make it exceedingly easy to get started with intent classification – without having expertise in NLP or machine learning. MonkeyLearn works behind the scenes doing all the heavy lifting for any person interested in intent classification with AI.

To get started, we recommend that you use our pre-trained model for intent classification that categorizes outbound sales email responses based on subject and body. In this pre-trained model, you can type a sample email response and the model will automatically identify different intents which include Autoresponder, Email bounce, Interested, Not interested, Unsubscribe, and Wrong person. It’s a good way to test out, with your own text, just how simple it is to use a machine learning-based intent classifier.

Once you understand how to use a text classifier with MonkeyLearn, we suggest creating a custom model. By creating your own model using your own data, and defining intents tailored to your specific needs and business, you’ll be able to ensure higher accuracy when identifying customer intent.

It’s super easy to create your own model for intent classification! Follow these simple steps below, and you’ll be ready to pinpoint intent in no time:

1. Create Your Classifier

First, you’ll need to sign up to MonkeyLearn for free. Then, go to dashboard and click on ‘create a model’. This action prompts you to choose a model type. For this tutorial, choose ‘classifier’.

2. Select the Classification Type

There are three options: topic classification, sentiment analysis, and intent classification. Choose ‘intent classification’.

3. Import Data

Upload the data set you want to use for training your intent classifier. You can upload data using an Excel or CSV file, or one of our many integrations.

4. Define Tags

After successfully importing your data, you can begin to create tags for your intent classifier model. You need at least two to train your model – bear in mind that the more tags you add, the more training samples you’ll need to train your model:

Tag Data to Train the Classifier

As you tag text data, the model will learn from your examples and criteria, and its prediction level will increase – remember, the more training, the more accurate your model will be.

Test and Evaluate the Classifier

Once you’ve trained your intent classifier, you can test it by going to the ‘run’ tab and typing a sentence into the text box, then clicking ‘classify text’ so your model can analyze and make predictions:

Put the Model to Work

Once you’re satisfied with your model’s predictions, it’s time to analyze your data. There are three ways to upload and analyze your data:

  1. Process data in a batch by importing an Excel or CSV file
  2. Use one of the available integrations (e.g. Google Sheets, Zapier, Zendesk, or Rapidminer) to connect with MonkeyLearn. 
  3. Or if you know how to code, you can use MonkeyLearn’s API to analyze data programmatically.

Wrap-up

Intent classification can be your best ally when it comes to transforming leads into customers. By using AI to your advantage, you can analyze massive amounts of interactions from your users and potential customers, and have a machine learning algorithm automatically detect the intent behind each one.

As you automate this task, you can immediately take action and contact qualified leads in a timely manner. And the best thing – it is very easy to get started with intent classification with AI. Just sign up to MonkeyLearn for free and start using pre-trained models right away. If you need help, don’t worry, just request a demo and our team will help you get started with intent classification.

Federico Pascual

Federico Pascual

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

Notification

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.