Chatbots: More than just customer service tools

For a while, chatbots where the next big thing in the customer service segment. However, as the "hype" around the technology is slowing down, its influence grows beyond just one business case.

Content: 

1. What is a chatbot? 
2. Does every chatbot have to be intelligent? 
3. AI personalizes and expands chatbot functionalities
4. Language is at the core of successful conversation
5. So, what does this mean for chatbot use in your company?

When we talk about chatbots, most people have a chat window in mind and think of rather linear conversations for sales or service purposes. But chatbots are and can be much more. Already, they are part of our lives in many different ways and with AI technology gaining traction, they can revolutionize the way we do things, handle machines and gadgets, and gain information both in digital and analog environments.

First things first: What is a chatbot?

Back to start of page

A "chatbot" describes any software used to have an online chat conversation. Although most people immediately think of chat windows, a chatbot can also be implemented on the phone or as part of another software such as a GPS or media system.

In general, a chatbot replaces or accompanies a human agent or enhances an experience that usually does not involve service options.

There are different ways of designing chatbots. Some are programmed with very specific conversation directions, meaning that the user usually has to say/write specific keywords to receive the right reply. These types are usually called "transactional" since they have one very clear task.

However, more sophisticated chatbots such as Siri or Alexa also learn with each interaction by using artificial intelligence and machine learning. They are what we know as "conversational chatbots" since they are able to hold and sometimes even maintain a conversation with the user.

Does every chatbot have to be intelligent?

Back to start of page

Now, there are possibilities to create simple chatbots by working with decision trees and predefined answer-question-dialogs. Not every chatbot needs to be a self-learning Siri that gets smarter with every single interaction.

These simple chatbot versions have their functions and advantages, especially as a first entry point to provide frequently asked questions (e.g., opening times, addresses, contact info, availability). As such, they are often used on social media channels and on Homepages and help alleviate the workload of the live agents by filtering out easy-to-solve problems from issues and questions that demand human support and expertise.

These chatbots are also perfect to gather information before a live agent takes over, such as name, customer ID, problem, etc. This way, the data gets collected and the live agent (whether for marketing, sales or services) can step into the conversation with a clear view on the issue which in turn optimizes the overall experience.

AI personalizes and expands chatbot functionalities

Back to start of page

Now, AI-based chatbots can deliver more complex customer experiences and so much more. Both chat and voice assistants have been increasingly taken over our daily lives. Whether it's asking Siri to make an appointment, listening to your GPS system how to avoid heavy traffic or working your way through a smart banking chatbot to get a new credit card - nearly everyone has interacted with a chatbot one way or the other in the last couple of years.

• With further development regarding machine learning and faster ways to analyze data, chatbots have grown their capabilities and use cases.
• They can train and guide users through complex processes. Used internally, they can also enable and educate employees and partners.
• They can help with the maintenance and handling of machines and guide interactions with different devices to reduce risks but also help people with disabilities navigate their private and work life.
• They can optimize internal requests and tasks such as the managing of budgets, vacations, IT issues, etc.

Our DIGITALL experts are happy to support you in setting up AI foundations and developing chatbots. Additionally, we can help you evaluate solutions to create simple chatbot workflows, for example with Microsoft Power Platform.

Language is at the core of successful conversation

Back to start of page

One of the key drivers of chatbot technology lies within NLP, short for Natural Language Processing. NLP covers AI technology covering numerous topics that are all closely related to understanding and processing written and spoken language. The field is vast, since it covers things like:

• Understanding written and spoken language including typos, synonyms, accents, speech impediments, slangs, etc.
• Processing (aka understanding) the text meaning, so the AI can reply properly with the right solutions/answer
• Formulating a proper answer in a cohesive, understandable text
• Translate the text into different languages (or into different formats, e.g., for a screen reader)

NLP is both one of the most promising and complex fields of artificial intelligence in general, since it is even complex as a human to always fully understand, process, and respond to a message. It is also crucial for people to understand this complexity when they want to incorporate smart chatbots into their user/customer experience. Otherwise, it's easy to set the goals for the chatbot too high to successfully fulfill them.

No matter how smart the bot, precise use cases with clear goals are still the best way to approach the development instead of trying to create a bot that can do it all.

Even seemingly all-knowing smart assistants like Alexa have clear underlying functions: answering questions (by using Google search functions), performing simple actions (such as making appointments or writing text messages), and having simple conversations (answering to easy questions or telling a joke).

As such, these assistants work mainly with

• keywords (selecting results that feature certain keywords in combination such as "food" and "delivery"),
• probability (how probable is it that someone who uses the word "weather" wants to know the weather of their current location?),
• individual preferences & habits (such as location, usual searches, activities, etc.),
• and simple actions (input of information in the calendar, as a text message, etc.).

So, what does this mean for chatbot use in your company?

Back to start of page

If you want to develop and/or implement chatbots internally or externally, make sure to define proper use cases that have a clear function and goal (e.g., be a first information point for homepage visitors or collecting contact data for marketing, sales or service).

Try to identify problem areas in your customer/user interaction that can easily be optimized via chatbots by creating shortcuts, give information or automate simple workflows.

Consider the type of chatbot you want and need. Whether you can work with a chatbot that uses so-called decision trees that are predefined or whether you want to make use of artificial intelligence, so the chatbot can learn and adapt.

Make sure to allocate the right resources to your chatbot project. Do you have sufficient expertise within your company or do you need external help? Do you have the basis (e.g., processing power & data base) for AI technology?

Define goals and key performance indicators that you can measure to see how the chatbot performs and is used. Analyze the results regularly to see how users adapt the new technology and how satisfied they are with its services.

Optimize and change if necessary. One of the reasons why smart chatbots are so complex and even controversial is the fact that developers might have one specific use in mind that the actual users not necessarily agree with.

One of my favorite examples of intent vs. result of smart assistants is chatbot Timmy by the airline Condor (source: horizont.net). Initially, they developed Timmy to perform up- and cross-sell services by selling additional airline perks to its users. However, the users activated Timmy predominantly to ask about flight information. Instead of considering Timmy a failure, Condor did the logical thing and adapted Timmy to meet the customers' requirements.

Find out more about digital megatrends, success factors of the digital transformation and best practices on leadership and company culture. Subscribe to our blog newsletter.