What makes a great chatbot?

TOP 8 skills users expect from a chatbot 

“What makes a great chatbot?”

The ultimate question that moves the chatbot industry, without having a clear answer for. This question is so difficult to answer because there are so many variables to consider when working on chatbot development and conversational design.

That’s why we have taken on the challenge and teamed up with Artificial Solutions, the company behind Teneo, to provide you with eight of the best features a chatbot should have and why they are so important to include in the development of a Conversational AI solution.

Functional chatbot features 

Our minds are hardwired to make some situations more complicated than they are. It is easy to come across articles focusing on how to “add personality to our chatbot”, yet even the simplest user example can break them.

 “Simplicity is the ultimate sophistication”

No matter how much we want to believe that the industry has reached a stage where only soft touches can play a differentiator, we must face the reality that we also experience when we try to communicate with a chatbot as a user.

What we know for sure from market research is that functional features are undoubtedly the most important for users. Functional features are responsible for a product or service to reach its primary objective. We list some features here that directly serve the chatbot to fulfil the task at hand.

 

Accurate chatbot answers

While choosing an AI chatbot building platform, invest time in research so that you are certain that the chatbot can provide accurate answers. We also often find ourselves disputing about which one is the best chatbot building platform. The answer is different for each industry depending on a lot of factors. Every platform has unique capabilities that work well for a certain industry but does not perform so great in another business case.

For example, with Teneo you don’t have to make the choice between using machine learning (ML) or linguistic understanding (LU) as the platforms uses a hybrid approach. Within the same solution, you can start quickly getting great coverage with ML triggers and then layer in linguistic understanding where required to obtain much greater precision and reliability than could ever be achieved with ML alone. They can even be combined within the same flow or even the same trigger condition Botium Box includes functionality to evaluate the performance of your training data on supported platforms and output relevant quality metrics.

 

 

Before making any commitment to a chatbot technology, it is worth doing such benchmark analysis to compare the outcomes and choose the platform with the most promising result for your use case. Keep in mind that the clear precondition for a fair comparison is for these bots to be trained with the same training data.

 

Chatbot context understanding

Context-awareness endows chatbots with the capability of interacting with users in an efficient, intelligent, and natural way.

These chatbots remember the things a user has already asked or performed and based on that, it presents a more thoughtful answer. For instance, you asked the chatbot about the weather and let us assume you also received an adequate answer. After that you ask: “And what about tomorrow?”

The chatbot must understand the context and instead of expecting the user to repeat the question like “What will the weather be like tomorrow?” give users the option to have a more natural conversation with a bot. Nearly all chatbot engines provide the possibility of context-tracking and despite the high efforts it is really something worth implementing!

To help with this, Teneo has out-of-the-box Conversational Modules that boost your bot with lots of pre-built intents and flows, allowing it to understand and respond to greetings, meta-requests (like Can you repeat that?), feedback, conversational small talk and other conversational topics.

 

Ability to fail usefully

Since people can say or write anything to a chatbot, it is an unrealistic expectation that your chatbot will be able to respond wisely to every user example. They can break, misunderstand, and make errors. But what is important is how those failures are dealt with. A chatbot failure, though not ideal, does not always need to mean bad user experience.

Make sure that the users are informed about the failure, so they can decide what further action to take. This should also result in supplying options in times of chatbot failure. After an occurring fallback intent, make it easy for users to abandon the failing bot and connect with a human agent.

Prevent as many failures as possible and learn the reason behind these crashes. By using Botium, you can monitor your Teneo built chatbot and receive notifications about failures in production.

 

Availability on multiple platforms

Businesses are always trying to cater to the users, so that they do not need to familiarise themselves with a new chatting environment.

Regardless of what platform the customer is using, they should always have a consistent user experience that also embraces brand recognition and trust. On Website, Mobile application, SMS or even on physical kiosks, they should talk to the same conversational AI and receive the same level of service.

To provide such persistent service, you need to be able to test the chatbot on each interface as well. Botium offers E2E Testing on all channels and platforms. Social messengers, websites, mobile apps, voice applications and IVR (Interactive Voice Response) systems.

Multiple channel availability enables you to engage with your customers via a platform that they are already familiar with. Teneo currently has connectors created to make solutions available over 20 different channels.

 

Part 2 – Non-functional chatbot features

We are delighted that you have made it this far! Now, head on over to Artificial Solutions’ blog to discover the next four features that you MUST include in your chatbot development project!

Read Part 2, here!