Everything you need to know about chatbot support

AI technology can enhance your brand’s digital interactions, improving the overall customer experience. We’ll explore how your brand can effectively implement chatbot support as well as other uses for bots beyond chat.

“A chatbot is a conversational piece of software powered by pre-programmed responses or artificial intelligence (AI) to answer questions without the need of a human operator,” writes G2. There are rule-based chatbots, AI chatbots, and machine learning chatbots. Chatbots can live within any major chat product, like Facebook Messenger, Slack, Telegram, text messages, or a brand-owned messaging channel like Khoros Messaging.

Chatbot customer service is a big opportunity for brands because people are now using messenger apps more than they are using social networks, writes Chatbots Magazine. Chatbots offer unique benefits, like the ability to offer 24/7 service, and they can easily acquire customer information, writes G2.

Our ebook, The Bot Balancing Act, explains that AI in general and chatbots in particular are receiving a lot of attention now because many brands need to handle ever-increasing volumes of digital communications. But, with the enthusiasm comes concern: Brands fear negative impacts to the customer experience with chatbot support. However, your brand can meet those concerns head-on and overcome them by following the best practices outlined in our ebook, along with those detailed below. You can find even more information about bots in our ebook, The Bot Balancing Act.

Chatbots are powered by either pre-programmed responses, artificial intelligence, or a combination of the two. Chatbots work on an applied mechanism where a chatbot will process a user's query to deliver a matching answer. There are two main types of chatbots, rule-based and AI.

Rule-based chatbots

Rule-Based chatbots, also called transactional, communicate with predefined answers. They have a script and they will stick to it. Most chatbots operate under a set of if/then rules that can have varying degrees of complexity. These chatbots are similar to a pet dog, they understand certain keywords and how to respond to those, but they will not understand the context of a conversation or understand the complexities of human language. If it is not a command they have been taught, they will not be able to help a user.

If you ask a chatbot a question such as “How do I track my order? '' First, the chatbot will look for familiar keywords in the sentence. In the example provided, it will take out words like ‘track’ and ‘order’. From there it will match these keywords with available responses in its database that have answers to questions involving these keywords. Rule-based chatbots often have problems detecting questions and keywords outside of the scope of available data to the chatbot, say the user has different spellings such as the British ‘colour’ vs. American ‘color’. Different dialects may also present the same issue as different spellings; either of these may result in the chatbot asking the user to rephrase the question or simply transferring to a human user.

While rule-based bots have less flexible conversational flow and lack an understanding of the complex language they do have their advantages over an AI chatbot.

• Rule-based chatbots have more predictable results when compared to that AI chatbots.

• Generally faster and less expensive to train.

• Integrate well with legacy systems.

• Can include interactive elements and media.

• Are not restricted to text interactions.

• Accountable and secure method of user management.

• Can streamline the handover to a human agent.

AI-based chatbots

AI bots are a more sophisticated version of rule-based chatbots. These bots tend to work well for companies that have a lot of data on their interactions. Initially, they will take longer to train, however in the long run they can save time as they get better and more advanced in problem resolutions. AI chatbots are much better conversationalists than a rule-based system because they have machine learning, natural language processing, and sentiment analysis.

• Machine learning

Machine learning allows chatbots to identify patterns in user inputs, decision making and allows the bot to learn from past conversations. Initially, your bot may seem less advanced than a rule-based variety, but very quickly AI chatbots will soon surpass the rule-based cousins in almost every aspect of problem resolution.

• Natural language processing

Natural language processing (NLP) allows AI chatbots to understand the complexities of human communication and enables them to replicate that level of communication. Natural language processing allows AI chatbots to understand the context of the conversation, regardless of dialect, or typos made by the customer. This often leads to less frustration associated with communicating with the more one-dimensional rule-based chatbots.

• Sentiment Analysis

Sentiment analysis allows chatbots to understand a user's emotions. With this ability, your chatbot can learn more effective ways of communicating that can lead to happier customers. If your chatbot can more successfully lead to positive interactions, you can increase brand loyalty and retention. When chatbots detect negative sentiment, they can use priority routing to get the customer in front of a human agent as fast as possible. Being able to detect your customer’s emotions can be a beneficial factor in all aspects of the user’s experience.

Along with the above points, AI chatbots have some other benefits of rule-based chatbots. AI chatbots:

• Will continue to learn from the gathered information.

• Continuously improve as more data comes in.

• Can understand multiple languages.

• Have a broader range of decision-making capabilities.

• Understand patterns of behaviors.

To effectively use chatbots for support, brands should balance business efficiency with customer experience. As we write in our ebook, “In chatbot use cases, the bot interacts directly with the customer. The bot may only engage for part of the conversation — for example, with a welcome message and/or expected wait time for an answer — or the bot may handle the entire conversation. The latter scenario is often referred to as a bot contained conversation.” Other chatbot support use-cases that also involve customer care agents include welcome messages, simple process automation, and simple problem resolution.

Two examples of effective chatbot support:

Chatbots Magazine details an example of how chatbots could theoretically help customers place orders online — in this case, with Nordstrom. Rather than searching on the website for a pair of shoes and then placing an order, customers would instead go to Nordstrom’s Facebook page and send a message to the brand describing the pair of shoes they’re looking for and a chatbot would mirror “the type of experience you would get when you go into the retail store.” In this case, the chatbot functions like a personal (digital) shopper. Not only will they improve the customer’s shopping experience, but an AI chatbot digital shopper will sell your products and services for you. They can provide a more personal experience to your customers while also gaining valuable customer insights.

In our ebook we write about a real-world example of chatbot customer service from a top-tier telecom provider. The provider, “decided to offer messaging in its mobile app (via Khoros Brand Messenger) to enhance customer experience. But, they knew inbound volume would surge as a result.” To manage, the provider started small: they initially limited in-app messaging to a select set of customers, rather than opening it up to their entire customer base. The chatbot greeted customers with a welcome message. The chatbot was powered by machine learning and was connected to the provider’s existing support community, from which it could find answers. As customers engaged with the chatbot, it learned quickly: In just 30 days the chatbot was handling one-third of inbound service requests, all without the aid of a live agent. Giving your AI-based chatbot support system time to ramp up can be just as important as implementing it. 30 days for a one-third reduction in inbound service requests really shows the power and scale of what an AI chabot support system can offer.

Wherever your brand is on its chatbot customer service journey, the following best practices, adapted from our ebook and G2, can guide your efforts and ensure you’re offering the best possible customer experience:

1. Use human support when needed. Chatbots aren’t fool-proof, and often a real person will need to step in to assist customers. Chatbot support is often used to get the initial parts of a customer support interaction completed before passing off those customers to a human support operative. This will increase efficiency and decrease your customer support lines.
2. Be upfront with your customers when they’re talking to a bot. Make sure customers know when they’re chatting with a bot and when they’re chatting with a person. You can simply name the bot and the person, or you can mark communications from both with clear icons (a robot for the bot, for example). Nothing can be quite as frustrating as when you think you are communicating with a person, only to find out the scope of their capabilities to assist you are limited to a rule-based chatbot.
   
3. Keep communications from the bot concise and short. Two to three sentences are preferred for most situations as quick, efficient communications are one of the main benefits of chatbots. People do not want to have lengthy conversations with a chatbot, they simply want to know how to resolve their issues. The benefits of a Chabot include being able to be very direct with your solutions.
4. Prioritize in-channel resolution. This will ensure a better customer experience. Your customers do not want to be bounced around during their support. Being able to get the assistance they need in one stop will make that experience much better for them.
5. Gather subject matter experts from different teams within your organization. Expanding your chatbot’s capabilities from different facets of your business will offer the most support to your customers. If you are investing in a chatbot, there is no reason not to have them capable of assisting each department in customer support.
6. Plan beyond your current bot. Simply put, things change, especially with AI technology. Choose a framework that evolves. This will give you a higher return on your investment. Having a chatbot that learns means it will get better over time. Think of chatbots as a new employee, if you constantly are bearing the onboarding cost, your efficiency decreases. Getting a bot that can learn and grow with your company will provide the most lifetime value, and have much less downtime than if you are constantly having to update a rule based chatbot support system.

Bots (or AI) can be used to improve the customer experience in other ways beyond direct chat. They can also be used as triage: they can help agents behind the scenes in resolving issues quickly, leaving agents free to engage in higher-level customer interactions.

Another valuable non-chat use for bots, as we write in our ebook, is gaining insights. More specifically, bots can help “filter out the signal from the noise with inbound posts, isolating and presenting to agents only those posts most likely to require agent action.” Khoros Service can assist your brand in processing large amounts of data and can offer your brand valuable insights about patterns over time.

Download our ebook to learn more proven ways to use bots for customer support.