eBook

Straight answers to the 10 most important questions about getting started with chatbots

Know Before You Bot

Next Chapter Download PDF

Chapter 3

What type of bot do I need?

Depending on your use case, the bot’s scripts, functions, and even relationships with your agents will vary. Chatbots come in many different flavors – from text to button, from keyword to AI-powered language recognition. Each type has its own unique set of capabilities based on the different AI chatbot architectures. You need to understand their differences before you start building your bot.

Start by establishing some must-haves. Does your chatbot need to answer a specific set of queries, or does it need to be open to a broader range? Does it need to be able to provide content to the customer (e.g. images, article links, videos, product reviews)? Does it require access to dynamic information (e.g. product database, current exchange rates, daily weather forecasts)?

There are basically two ways for your customers to interact with a chatbot:

  1. Typing: Letting customers interact with your bot through their own words gives them freedom to phrase questions in different ways. But it’s give-and-take; this more ‘open’ interface may feel more natural to the customer, but it requires a lot more training and maintenance by you.
  2. Buttons: This interface lets you control the scope of your bot experience. Customers are limited to pre-defined choices along a fully mapped out ‘decision tree’, clicking on buttons instead of typing to get the answer they seek. The trade-off here: a safer and more controlled experience for you, but a narrower range of capabilities for your customers.

If you want your customers to be able to interact through typing as they would with a human agent, then you now have to choose between two basic types of language interface:

  1. Keyword: These chatbots can be programmed to provide readily accessible answers to straightforward questions. This type of language bot identifies preprogrammed keywords or phrases used by your visitors, and serves up the associated responses.
  2. Natural Language Processing (NLP): Thanks to their ability to identify customer intent and to learn from their conversations, these chatbots offer the most natural interface for your customers. Their use of AI equips them to handle the broadest range of questions with less programming and maintenance.

Note that both text and button-based bots can integrate with other systems in your customer experience tech stack, so it really comes down to your desired scope and preferred type of interaction. You may end up with a healthy mix of all types of bots.

Consider this

Ask your chatbot vendor what types of bots they offer, and which would be best for your use case. Ideally, find a vendor that offers all types of bots that can meet your requirements, rather than selling you something too advanced, or not advanced enough.