Intent Detection: Unlocking The Full Potential Of The Chatbot Experience

Conversational Assistants are an amazing tool that teams can leverage to create chatbox experiences that engage and interact with clients. They can be designed to deliver information, collect data, automate processes, or even simply offer a welcome greeting before an agent takes over.

Assistants can be configured for different text-based channels, including Chat and SMS, and are constructed out of interactions, grouped by dialog. Dialogs are composed of interactions, strung together into a complete chatbot experience. They can even branch out conditionally into different paths, based on a customer’s responses.

This baseline experience is built to guide customers to the information they need, provided they know how to navigate the experience properly. However, sometimes customers can end up in the wrong section of Assistant and need redirection. They may also want to skip the chatbot experience entirely, and get straight to a support agent as fast as possible. Being stuck in the wrong interaction, or not being able to have needs quickly met, can be a frustrating experience for any customer.

In an effort to optimize chatbot performance and avoid negative experiences like these, Intents can be configured to easily detect and understand common inquiries. Detecting an intent allows a chatbot to efficiently route a customer to the interaction that is best suited for their needs.

Say, for example, a customer wants to make a return, but they end up in an interaction for technical support, a different team. By detecting key words, the Assistant can quickly reroute the customer to the correct dialog and start the returns process.


To implement this functionality into a new or existing Assistant, you will need to define your intents through a collection of keywords and phrases by visiting Settings > Kustomer IQ > Conversational Assistants > Intents and selecting the option to add an intent.

An intent should be designed around a recognizable action or objective, for example, making a return. An intent can be designed to recognize up to 15 keywords (per language), and at least 5 phrases.

English keywords for a return-based intent might include terms like “return”, “refund” “defective” or “exchange”.

Keep in mind that a single keyword can only ever be used in one unique intent. Using the keyword “exchange” for the Return intent, for example, means you won’t be able to use it later as a keyword for a different intent. This ensures there will be no confusion when a keyword is detected during a chatbot encounter.

Phrases can also be provided as training for an assistant, to help a chatbot understand how a customer might word their inquiry in different ways. 5 phrases are required for training to training commence, but 12 or more are recommended.



Once the keywords and phrases have been established for your intent, you can implement intent detection from within a Conversational Assistant. You can set a dialog to listen for intent on any Customer Text Response type interaction, or any other interaction showing the robot icon.

To listen for intent in a customer’s response, simply activate the toggle—

From here, you can configure any dialog to trigger based on the intent that was detected!



Say you want to allow customers to request immediate help from an agent, instead of going through the entire chatbot experience first.

  1. Think about the various ways in which your clients might express this need, and come up with a list of keywords and phrases to base the creation of an intent around.

  1. Set up an Assistant dialog initiates Human Takeover to trigger when this intent is detected.

  1. Toggle on intent detection somewhere (in at least one interaction) within the Assistant flow. It’s considered a best practice to use intent detection early and often!

  2. Make sure to test your new chatbot experience to make sure it works effectively and as intended!

Implement intent detection to add even greater versatility to your automated chatbot experience, and help get your clients what they need with speed and efficiency!

To learn more about Conversational Assistants and using intents, check out these additional resources:

Article: Introduction To Conversational Assistants

Article: Conversational Assistant Dialogs

Article: Detecting Intent With Conversational Assistants

Article: Conversational Assistant Rules

Kustomer University Course: Conversational Assistant Fundamentals

Kustomer University Course: Advanced Conversational Assistants

Kustomer University Video: Getting Started With Chatbots