Welcome to our conversational guide to the customer and business intents, which will help you understand, classify, match, and use these fascinating concepts in conversation design and dialogue management. Feel free to read the contents of this guide in order or jump straight to the section that sparks your interest. Here's a list of topics covered in this guide:
- Definition of Intent
- Definition of Customer Intent
- Types of Customer Intents
- Ways to Classify Customer Intents
- Definition of Customer Intent Matching
- How Customer Intent Matching Works
- Ways to Match Different Types of Customer Intents
- Difference Between Customer Intent Matching and Classification
- Definitions of Business Intents
- Goals of Business Intents
- Ways to Match Business Intents
- Conclusion
What’s an Intent?
An intent is a fundamental concept in conversation design and dialogue management. It signals to conversational AI that either a customer or a business wants to take the initiative in the conversation to satisfy relevant needs.
What’s a Customer Intent?
Customer intent provides an insight into why the consumer engaged in a conversation and describes the information they intend to find or the task they aim to accomplish when they speak or type their request as an utterances to a conversational bot.
What Are the Types of Customer Intents?
There are various reasons why potential or existing customers engage in a conversation, and the following types of customer intents aim to classify them into four distinct categories:
- Informational Customer Intent
- Commercial Customer Intent
- Transactional Customer Intent
- Navigational Customer Intent
What’s Informational Customer Intent?
Informational customer intent is exploratory in nature. Consumers with this type of intent might not be familiar with the subject and usually have only a high-level understanding, so they might not even use the correct terminology when they phrase informational questions. They seek reputable sources of information that will guide them through acquiring new knowledge that will help them satisfy their interests.
What’s Commercial Customer Intent?
Buyers express commercial customer intent after satisfying their initial informational needs and have decided how they intend to solve their problem. They look for ways to compare the available solutions before moving to the transaction stage. They perform comparative research by asking commercially focused questions to ensure they make the best possible buying decision to yield the most benefits in their particular situation.
What’s Transactional Customer Intent?
Contrary to common belief, transactional customer intent isn't only about accomplishing a purchase. It's related to any task or activity a consumer wants to perform, such as finding a place that offers specific products or services in a particular location, checking availability, booking, rescheduling or cancelling service or product delivery.
What’s Navigational Customer Intent?
Potential or existing buyers with navigational customer intent know what they are looking for and want to find relevant information, such as how to get to a physical location or get in touch with a specific person or department of an organisation. They aren't looking for general guidance but ask navigational questions in a way that aims to find exact details that will help them fulfil their need.
How to Classify Customer Intents?
Consumers provide all the necessary information about their intent in the utterances they used to phrase their query, so classifying customer intent involves correctly applying relevant rules for each one of them, which is possible by using one of the following methods:
What’s Manual Customer Intent Classification?
Manual customer intent classification is a term used to describe the intent classification process performed by people. It requires human attention to classify an intent into one of the four categories by looking at the utterances and writing the appropriate type of customer intent next to it.
How to Classify Customer Intents Manually?
The manual method for classifying customer intent relies on applying rules based on the presence of certain words and their relations that convey meaning and indicate informational, commercial, transactional or navigational consumer intention. These words are sometimes called intent modifiers.
How to Manually Classify Informational Customer Intent?
Potential or existing customers with informational intent ask questions and often include the following words in their utterancesŸ: "who", "what", "why", and "how", which indicate their aim to improve general knowledge in a particular field of interest.
How to Manually Classify Commercial Customer Intent?
Consumers with commercial intent ask questions about the best products or services, the pros and cons of one solution versus the other, the provider’s trustworthiness and expertise, prices, locations, and delivery options. They often include words such as “best in class”, “versus”, “top”, and “cheap”, which indicate their aim to compare available options that suit their needs.
How to Manually Classify Transactional Customer Intent?
Instead of asking questions, buyers with transactional intent phrase their utterances as affirmative sentences. Words used by consumers with this type of intent indicate an action they wish to perform, such as "book", "reschedule", "cancel", "purchase", and "buy" and often include their area of interest either by providing a specific location or by using phrases like "near me".
How to Manually Classify Navigational Customer Intent?
Customers with navigational intent ask questions including brand names in combination with the following words in their utterances: "how", "when", and "where", which indicate that they want to find a way to contact a business representative, get to the business location or want to ensure that they get there on time.
How Effective is Manual Customer Intent Classification?
Manually labelling utterances with customer intent categories is a labour-intensive task. However, a trained person who follows manual intent classification rules can deliver top-quality results unachievable for automated intent classification solutions.
What Are The Advantages of Manual Customer Intent Classification?
While manual intent classification isn’t free from limitations, it has several advantages over automatic classification, including the quality of the results, which are attainable even for people without technical expertise.
Why Is the Quality of Results One of the Main Advantages of Manual Customer Intent Classification?
Although some utterances have overlapping customer intent types and can be hard to classify even for people with lots of expertise, the manual intent classification process yields the best results.
Why Is No Need for Technical Expertise One of the Main Advantages of Manual Customer Intent Classification?
Automated intent classification requires significant technical expertise, but the manual process relies on simple rules that everyone can quickly understand and apply. Since it doesn't require a lengthy learning process, it shortens the time necessary to take advantage of this knowledge and enables more companies to see the benefits of improved messaging.
What Are The Downsides of Manual Customer Intent Classification?
While manual intent classification has advantages, it's not free from limitations. Some of the crucial downsides include the laboriousness of the process, which is difficult to scale with the demand and makes it impossible to apply automatic updates.
Why Is Laboriousness One of the Main Disadvantages of Manual Customer Intent Classification?
Since humans do all the work, manual intent classification requires the human classifier to acquire the necessary know-how and takes their time to look at every entry in the list and write its class next to it. While it might not be a problem for a small number of utterances when it grows beyond staff capacity, switching to an automated process becomes necessary.
Why Is Difficult Scalability of Manual Customer Intent Classification One of Its Main Downsides?
Manual intent classification requires human attention, which is challenging to scale and can sometimes cause problems when the demand for these activities changes. It results in either staff underutilisation or overload. The former generates fixed costs without yielding enough value for the business, which indicates inefficient resource management, and the latter can lead to employee dissatisfaction and even burnout.
Why Is No Automatic Optimisation One of the Main Downsides of the Manual Customer Intent Classification Process?
With the growing experience, the person who manually classifies customer intents will produce better results. However, this kind of improvement might be risky for the business since the know-how acquired by that person is difficult to share and can be hard to quickly replace in case of that person's sickness or when they decide to work for another employer.
What’s Automatic Customer Intent Classification?
As the name suggests, automatic customer intent classification automates the manual intent classification process. After the initial training phase, an intent classifier can automatically process natural language and classify each utterances into one of four categories without human attention.
How to Classify Customer Intents Automatically with Conversational AI?
Conversational AI is well equipped to automate the laborious process of manual customer intent classification. It relies on AutoML to build machine learning models that process natural language to classify customer intents into four categories. Here's a list of steps involved in this process:
How to Preprocess Data for Automatic Customer Intent Classification with Conversational AI?
Humans can easily comprehend written language, but machine learning models used for automatic intent classification rely on numbers instead of words. Hence, the prerequisite for training an intent classifier model is text vectorisation.
What’s Text Vectorisation?
This process is responsible for converting written language into mathematical representations without losing the relationships between words and phrases. The resulting arrays of numbers are called vectors, hence the name text vectorisation.
How to Train an Automatic Customer Intent Classifier?
Automatic intent classification requires training an intent classifier model with a list of customer intents classified manually to produce a probabilistic output for each of the four types of customer intents. This process resembles teaching by showing correct examples and, like with humans, involves a great deal of trial and error.
What’s an Intent Classification Algorithm?
An intent classification algorithm is a machine learning algorithm that contains step-by-step instructions enabling machines to learn how to classify customer intent from a list of correctly categorised examples and represent their learnings in an intent classification model.
What’s an Intent Classification Model?
An intent classifier model results from training an intent classification algorithm with a dataset that contains lists of manually classified intents, which act as role models. It possesses intelligence acquired during the training phase.
How to Predict Customer Intent Category Automatically with Conversational AI?
Once trained, an intent classifier model can automatically predict the type of customer intent on any number of new utterances and classify them into one of four categories of intents in real-time.
How Effective is Automatic Customer Intent Classification?
Intent classification models can be pretty precise, reaching up to 90% accuracy compared to manual classification. However, no matter how good the intent classifier model becomes, customer intent still has some ambiguity because even humans don’t always agree on classifying some utterances.
What Are The Advantages of Automatic Customer Intent Classification?
While automatic customer intent classification has some downsides, it has several significant advantages over manual classification, including its ability to deliver predictions in real-time, seamlessly scale with the demand, and apply automatic optimisations.
Why Are Real-time Predictions One of the Main Advantages of Automatic Customer Intent Classification?
Automatic intent classification can produce quality results in a fraction of the time needed for manual labelling of customer intents. It can be especially beneficial when the number of intents to classify is overwhelming for human capacity or when real-time predictions are required.
Why Is Instant Scalability One of the Main Advantages of Automatic Customer Intent Classification?
Automatic intent classification can seamlessly scale up and down in response to the actual needs. It's much more efficient than the manual process since it doesn't generate fixed costs when the demand is low and can instantly produce high throughput when it spikes.
Why Is Automatic Optimisation One of the Main Advantages of Automatic Customer Intent Classification?
Automatic optimisation allows for continuous improvement of automatic intent categorisation over time. In contrast to manual intent classification, which stays with the person performing the task even if they decide to work for someone else, it guarantees improved classification capabilities continue to be an asset for the company.
What Are The Downsides of Automatic Customer Intent Classification?
The main downside of automatic customer intent classification is that it requires significant technical expertise to preprocess the data, select an appropriate classification algorithm, train the intent classifier model, and use it to predict customer intent for new utterances.
What’s Customer Intent Matching?
Customer intent matching (also known as consumer intent recognition or detection) is a technology-powered ability used by conversational bots to automatically understand the meaning behind customer intent expressed as an utterances in real-time and match it with the best available response.
How Does Customer Intent Matching Work?
When clients express their intentions using natural language during a dialogue with a conversational bot, it needs to compare the meaning behind their utterances with a list of available intents to find the best match. There are two algorithms typically used to achieve this goal:
How Does Grammar-Based Customer Intent Matching Work?
Grammar-based customer intent matching relies on syntax analysis to identify parts of speech and relationships between them to generate insights into the meaning behind customer intents expressed as utterances. It uses this information to create search queries to match the most relevant documents stored in a knowledge base.
How Does Machine Learning Customer Intent Matching Work?
Matching customer intents with machine learning uses sample utterances as training phrases to generalise from them and generate a custom language model that can recognise multiple ways a potential or existing client might express the same intent to increase the chances of correctly identifying it.
What’s Automatic Expansion?
Automatic expansion relies on semantic analysis to generalise from sample utterances. It enables automatic recognition of many similar ways consumers express the same intentions without explicitly including them in phrases used to train the language model.
What Are the Pros and Cons of Grammar-Based vs Machine Learning Methods for Matching Customer Intents?
Grammar-based and machine learning methods for matching customer intents have pros and cons, making one of them preferable in scenarios where some aspects are more important than others. The following list covers all of the benefits and downsides of each one of them:
- Advantages of Grammar-Based Customer Intent Matching
- Disadvantages of Grammar-Based Customer Intent Matching
- Advantages of Machine Learning Customer Intent Matching
- Disadvantages of Machine Learning Customer Intent Matching
What Are the Advantages of Grammar-Based Customer Intent Matching?
While grammar-based customer intent matching isn’t free from limitations, it has several significant advantages over matching based on machine learning, such as:
Why Is Accuracy with a Small Number of Sample Utterances One of the Main Advantages of Grammar-Based Customer Intent Matching?
In contrast to intent matching based on machine learning, the grammar-based method doesn’t require a large number of sample utterances. Grammar-based customer intent matching delivers accurate matches even with few example phrases.
Why Are Quick Updates One of the Main Advantages of Grammar-Based Customer Intent Matching?
Unlike intent matching based on machine learning, the method relying on grammar rules doesn't require a long time to update the machine learning model. Any changes made to sample utterances that rely on grammar rules take immediate effect.
What Are the Downsides of Grammar-Based Customer Intent Matching?
While grammar-based customer intent matching has several essential advantages, it lacks automatic expansion, which can sometimes be a disadvantage compared to intent matching based on machine learning, which has this ability.
What Are the Advantages of Machine Learning Customer Intent Matching?
While matching customer intents based on machine learning isn’t free from limitations, its ability to perform an automatic expansion from sample utterances provides a significant advantage over the customer intent recognition based on grammar rules.
Why Is Automatic Expansion One of the Main Advantages of Customer Intent Matching Based on Machine Learning?
Potential or existing clients can phrase their requests in many different ways. Automatically expanding sample utterances can save time and improve intent matching accuracy, especially in situations where there are virtually infinite ways to express the same customer intent.
What Are the Downsides of Machine Learning Customer Intent Matching?
Although customer intent matching based on machine learning has an essential advantage over its grammar-based counterpart, it also has some downsides that can put it at a disadvantage when some of the following aspects are important:
- Requires Many Sample Utterances
- Slow Updates
- Matching Accuracy for Intents Containing Uncommon or Ambiguous Words
Why Is a Necessity to Provide Many Sample Utterances One of the Main Disadvantages of Customer Intent Matching Based on Machine Learning?
Automatic expansion is an excellent ability of intent recognition based on machine learning. However, training a language model relies on providing many sample utterances to enable effective generalisation, which takes time and human effort.
Why Are Slow Updates One of the Main Disadvantages of Customer Intent Matching Based on Machine Learning?
Training language models with automatic expansion can sometimes take much time, which puts customer intent matching based on machine learning at a disadvantage compared to the grammar-based approach when considering updates time efficiency.
Why Is Matching Accuracy for Intents Containing Uncommon or Ambiguous Words One of the Main Disadvantages of Customer Intent Matching Based on Machine Learning?
Matching customer intents in audio conversations relies on automatic speech recognition. One of its biggest challenges is recognising domain-specific vocabulary and ambiguous or difficult-to-pronounce words, which can result in poor matching accuracy of customer intents in these situations.
How to Use Customer Vocabulary to Improve Matching Accuracy for Customer Intents?
Creating custom vocabulary can effectively improve customer intent recognition, containing uncommon, domain-specific, difficult to pronounce or ambiguous words. The automatic speech recognition engine uses this vocabulary during transcription, which provides input for customer intent matching for audio conversations.
How to Match Different Types of Customer Intents?
Potential or existing clients initiate conversations to find a piece of relevant information or perform some task. While both of these activities sound similar, they usually require different methods for matching customer intent. The following list goes into greater detail about detecting each type of customer intent:
- Matching Informational Customer Intent
- Matching Commercial Customer Intent
- Matching Transactional Customer Intent
- Matching Navigational Customer Intent
How to Match Informational Customer Intent?
Since clients who express informational intent in their utterances look for the relevant information, they centre their questions around entities and their attributes, so a grammar-based method is well-suited to recognise this category of customer intent.
How to Match Commercial Customer Intent?
Because consumers express commercial intent by combining entities such as product, service and brand names with additional phrases to compare available options, grammar-based intent matching is typically a better choice for recognising this class of customer intent.
How to Match Transactional Customer Intent?
Since buyers express transactional intent by orienting their utterances around actions they wish to accomplish and phrase them in various ways, intent matching based on machine learning is usually the best approach to match this type of customer intent.
How to Match Navigational Customer Intent?
Because customers express navigational intent by combining entities such as product, service, brand or department names with phrases aimed at finding specific information related to time and location, a grammar-based intent matching is the right choice to match this class of intent.
What’s the Difference Between Customer Intent Matching and Classification?
Classifying customer intents is essential for creating tailored, conversational experiences that reflect different types of consumer intents. However, managing the flow of conversations and automating responses to these intents in real-time is another problem that requires matching customer intent instead of categorisation.
What Are Business Intents?
Business intents represent business goals a conversational bot accomplishes automatically on behalf of the company by taking the initiative at the right moment during natural conversations with potential or existing customers. The following list presents notable business intents:
- Welcome Intent
- Consumer Engagement Intent
- Sales Qualification Intent
- Reminder Intent
- Feedback Collection Intent
- Referral Collection Intent
What’s a Welcome Intent?
Welcome intent automatically greets potential or existing clients by returning one of the predefined welcome messages and briefly explains how a conversational bot can assist them and how they can start a meaningful conversation.
What’s a Consumer Engagement Intent?
Customer engagement intent automates the customer engagement process by engaging casual visitors in meaningful conversations presenting the value of products and services offered by the company in the context of customer needs.
What’s a Sales Qualification Intent?
Sales qualification intent automates the sales qualification process by prequalifying potential customers interested in business offerings by assessing customer needs and checking how well they align with business expectations.
What’s a Reminder Intent?
Reminder intent automates the laborious manual process of notifying customers about the date, time and location of their upcoming service appointments or product deliveries by sending messages over the relevant conversational channel ahead of time.
What’s a Feedback Collection Intent?
Feedback collection intent automates collecting opinions from customers shortly after their service appointments or product deliveries by asking them a few questions about the perceived quality of their experience.
What’s a Referral Collection Intent?
Referral collection intent automates collecting referrals from satisfied customers during natural conversations short after product or service deliveries by asking for contact details of people from the customer’s social circle who might benefit from similar products or services.
What Are the Goals of Business Intents?
The overarching goal of business intents is to convert casual visitors into loyal customers automatically. Conversational bots rely on business intents to move potential or existing customers to the next stage of the customer journey by implementing the following steps:
- Greeting a Potential or Existing Customer
- Presenting the Value of Products and Services
- Qualifying Prospects Interested in Purchasing Goods or Services
- Improving Timely Arrivals and Reducing No Show Rates
- Collecting Feedback After a Product or Service Delivery
- Asking Satisfied Customers for Referrals
What’s the Business Goal of Welcome Intent?
Every potential or existing customer is worthy of greeting them to make them feel welcome and appreciated. It sets a foundation for an upcoming business relationship. Welcome intent also provides an excellent opportunity to briefly introduce the company, its expertise, and revenue streams, which is especially useful for first-time visitors.
What’s the Business Goal of Consumer Engagement Intent?
Converting first-time visitors to paying clients starts with engaging them in meaningful conversations. Consumer engagement intent aims to show them the benefits of your business offerings in the context of their needs, which is an attempt to move them from the beginning of the customer journey to the next stage when they consider buying relevant goods or services.
What’s the Business Goal of Sales Qualification Intent?
Not every customer is the right fit for your business. Sales qualification intent prequalifies potential customers interested in your products or services to make sure you spend your precious time only on qualified prospects who are the right fit for your business, saving time and improving operational efficiency.
What’s the Business Goal of Reminder Intent?
Late arrivals and no shows can severely impact the efficiency of your operations. Luckily reminder intent can improve timely arrivals and reduce no show rates by sending reminders to relevant customers ahead of upcoming service appointments or product deliveries.
What’s the Business Goal of Feedback Collection Intent?
Collecting customer feedback is an essential part of assessing and improving your company's quality of services, but it can be a mundane and time-consuming task when performed manually. Feedback collection intent saves time and improves operational efficiency by automating gathering necessary insights for the incremental product or service improvements.
What’s the Business Goal of Referral Collection Intent?
Collecting referrals from delighted clients is one of the best ways to acquire new customers. It used to rely on human's ability to assess customer satisfaction before asking for referrals. Luckily, things changed, and referral collection intent can automate this process and ask only satisfied customers for details of other people who might benefit from similar products or services.
How to Match Business Intents?
Matching business intents relies on recognising customer intent and external events processed through rules defining when to trigger a particular business intent. The following list goes into greater detail about how to match each of the business intents:
- Matching Welcome Intent
- Matching Consumer Engagement Intent
- Matching Sales Qualification Intent
- Matching Reminder Intent
- Matching Feedback Collection Intent
- Matching Referral Collection Intent
How to Match Welcome Intent?
Matching welcome intent involves creating a combination of rules that trigger it in any of the following situations:
- A potential or existing customer initiates a conversation using standard greetings, such as “hi” or “hello”.
- An external event triggers it automatically when a consumer interacts through one of the integrated conversational channels.
How to Match Consumer Engagement Intent?
When potential customers express informational intent several times, they are broadly interested in the areas relevant to the business. While they might not be ready to make a purchase, it can be a great indicator to trigger consumer engagement intent to present them with the potential benefits.
How to Match Sales Qualification Intent?
When a business targets a specific market segment, for example, wealthy customers, it's worth matching sales qualification intent after the potential customer expresses interest in purchasing some products or services, but before fulfilling their transactional intent or transferring the conversation to a sales representative.
How to Match Reminder Intent?
Matching reminder intent relies on the following factors: the date and time of an upcoming appointment or product delivery and two rules defining time windows for sending the notifications. Usually, the first one is several days in advance and another one on the day.
How to Match Feedback Collection Intent?
Matching feedback collection intent relies on an explicit confirmation by the business representative of a successful product or service delivery followed by a confirmation from the customer or automatic confirmation when the lack of client acknowledgement exceeds a predefined time window.
How to Match Referral Collection Intent?
Matching referral collection intent relies on categorising customers as brand promoters. Analysing their feedback after the product or service delivery is a great way to derive this information automatically from their answers and increase the chances of successfully obtaining referrals.
Conclusion
An intent describes why a customer or a business might want to engage in a conversation. Classifying customer intents helps to understand the types of intentions that drive consumer conversations at different stages of the customer journey. Matching customer intents triggers relevant responses from conversational bots and assists in matching business intents that satisfy appropriate business goals.