When you get more than 1,000 inbound chats per month through live chat, it is time to think about automation. Until that point as a ballpark you get 33 chats/day. This is manageable for 1 person to handle.
Beyond that, however, you need some automation to reduce human workload in answering live chat. And additionally there are off-hours when agents are offline and customers cannot reach them.
For simple automation you can try the Zendesk Answer Bot. It suggests top matching articles based on what customers write either in email or in chat. It is a quick way to automatically increase self-service rates.
The requirement is that you have articles in Zendesk that the bot can pull up. You get up to 50 rule-based answer suggestions from articles in your Zendesk catalogue with Suite Team package €49/month.
Another option is to build a Conversational AI based chatbot (aka Intelligent Virtual Assistant) that can understand customer questions and provide conversational answers in the Zendesk chat widget. For this you need to select a third party chatbot provider and connect it to your Zendesk account (you need to be at least on the Suite Enterprise plan).
For starters you need to pick a vendor that has a Conversational AI based chatbot capability. What that means is that these intelligent agents are not only capable of presenting a multiple choice selection of answers to the user but also understand user intent from free text.
Understanding natural language is a significant leap. Many times users are looking to articulate their specific concern to the machine in a similar manner they would do to a human. User has a question and asks that specific question from the machine e.g. “When will I receive my payment from Bank ABC?”. The main drive behind this is that users are looking for a quickest way to get an answer to their specific question.
The conversational chatbot will be inside the Zendesk chat widget. It will be the first to respond to customer queries when they ask a question. When customers ask a question, the bot will match through machine learning the free text question to the best matching answer in its database.
If the bot does not know the answer, it will give users the choice (by presenting a button) to be connected to a live agent in Zendesk who will answer the customer question. The agent will see the conversation history between the bot and the customer.
Once you build the bot on the vendor’s platform you connect it to your Zendesk account. One bot takes up one Enterprise Suite license. If you want your bot to support many languages thenin most cases you need to buy several Enterprise Suite licenses (€150/month/agent) for your bots. There are some providers that provide multilinguality through one chatbot so you only need to purchase one Enterprise Suite license for your bot.
The main benefits for making customers happier are:
Yes. It is not easy of course and you will not achieve it on day 1 but it is doable. Some of the customers we work with have had such success. There are a couple of things that you need to do to set your bot up for such success.
First, pick the right answers to automate. Go for the answers that have the most volume. These get asked the most and because the bot works at scale you will have more chance at higher self-service rates if you focus on questions that get asked the most. So pick the 20% of questions that represent the 80% of volume.
Second, pick answers that you actually can automate. Ask whether the bot will have access to info that is needed to solve the question. Refer to the Topic Rating Matrix, it is an easy yet helpful exercise for finding the best topics for your chatbot to automate. For examples on how to authenticate, refer to this source if you are looking for ways to authenticate your customers through the chatbot.
Third, train your bot with quality data. More is not better. Higher quality training data is better. Make sure your topics in the bot are different enough and that your training phrases are closely related to the topic. Do not include personally identifiable data (names, ID numbers etc), this just confuses the bot.
Fourth, channel your customers to your bot. In your emails, contact pages direct people to the chatbot in Zendesk and instruct them to solve their issues there. That way you create the habit in people to seek help in the channel where you want to provide them help.
If you follow these principles then achieving 90% deflection rate through conversational chatbots in Zendesk is possible. These bots bring increased efficiency to your customer service and make customers happier through instant answers that are available 24/7.
If you are looking for AI chatbots in Zendesk for your customer service, then feel free to sign-up for an account with alphachat.ai In just a few minutes you can get your own natural language understanding AI chatbot ready with various industry templates that you can connect with Zendesk.