Source: Liam Pozz
Most consumers have noticed: there’s a huge gap in capabilities in the Chatbot industry. While one Chatbot may be able to book you a flight or serve you in many languages, another might not even seem capable of completing the single task that it were designed for, like telling you the weather.
What makes some Chatbots so good at solving issues, and others so hopeless? How can two Chatbots with the same range of abilities deliver two completely different customer experiences? In many cases, the difference in Chatbots performance is due to the quality of their decision trees.
Whether you’re designing a Chatbot yourself or outsourcing to a third party, it’s important to know what Chatbot decision trees are, and how you can use them to program your Chatbot successfully. This blog post will give you the run-down of everything you need to know about Chatbot decision trees before designing your Chatbot, and with tips to help you effectively plan and map your bot’s content.
Decision trees are how Chatbots help customers find exactly what they’re looking for: they map out a step-by-step process to discover the precise answer to the customer’s question in a conversational format.
The initial question is the “root” of the tree. From there, the Chatbot might need additional information in order to solve the issue, moving the conversation from general to specifics. For example, let’s say you’re using a fictional airline’s Chatbot, AirMaria. First, AirMaria might ask you what you want to do: book a flight, check an existing reservation, check arrival times, or check airline policies. You respond, “Book a flight,” and AirMaria moves you up a notch in the decision tree by asking you a new, more specific question: “What is your destination?” Once you answer that question, then the bot might ask you what dates, then times you are looking to fly (these are the branches of the tree). Finally, she may ask you to choose seats (these are the leaves).
The more specific an inquiry, the more branches will be needed to solve it. And the more branches you create, the more personalized and helpful your Chatbot will be.
Chatbot decision trees matter more than most people realize. As one Digital Product Agency writes, “Chatbots are only as good as the narrative itself, and storytelling is way more important than cutting edge tech or continuous content sheets.”
Decision trees define how Chatbots will handle each situation. Companies that don’t invest time and effort in their Chatbot’s journey mapping can wind up with dead-end bots, that hurt customers more than they help. Without a quality decision tree, the customer experience suffers.
For example, a Chatbot with a poorly designed decision tree might fail to ask a customer the right questions to lead them along the branches and towards resolution. It might fail to register certain key words, or only interpret limited patterns of human speech. Its branches might be too short, giving only general advice rather than specific, useful feedback – or they might be too long, taking the customer on a winding, never-ending journey to solve something simple.
Chatbots with well-designed decision trees take customers on smooth, coherent journeys towards resolution. Decision trees help even the simplest bots deliver on customer expectations by offering easily-accessible content that is tailored to different use-scenarios.
Decision trees can help customers solve complex questions. But in order for the Chatbot to deliver, the decision tree must be just as complex, and well executed. Here are some steps to creating a decision tree for your Chatbot:
Before you go into designing your decision tree, it has to be clear what customers and your company can expect from your bot. Is the bot supposed to help customers easily access simple support solutions in your knowledge base? Is it supposed to further your company’s marketing outreach by providing customers with a useful service, such as health tips, recipes, or investment recommendations? Or is it a “level two” Chatbot designed to perform more advanced functions, such as help customers change their passwords, or even place orders. How about all of the above?
Knowing where your bot will interact with customers will also help you decide what you can expect of it. Bots that interact with customers through your company’s app, Facebook Messenger, and SMS can send customers notifications and even daily content, whereas bots that assist customers on your website will likely have a stronger support function.
By being clear on what your Chatbot can and must accomplish – and on what platform – you will be able to start to design your decision tree with that in mind.
Once you know what to expect from your Chatbot, you can begin to establish some must-haves.
What information do users need to get out of your Chatbot in order for it to be useful? What queries must your Chatbot be able to answer? What kind of content does it need to push – i.e. product photos, article links, product reviews, etc.? What information does your Chatbot need to have access to so that it is always up to date – i.e. updated product database, current exchange rates, daily weather forecasts, etc.?
Some bots are quirky and fun, raining emojis and GIFs on their presumably millennial audience. Others convey a more professional image, keeping interactions simple and to the point.
When deciding on your Chatbot’s personality, ask yourself the following:
Asking yourself these questions will help you decide how your Chatbot will talk and act in every situation.
Designing the Chatbot flow is all about taking your customers through the process of using your Chatbot. How will you get your customers from Point A (their initial inquiry) to Point B (resolution, engagement, or any other endpoint, such as subscribing to your Chatbot’s service)?
There are many ways to guide the user’s journey. Some Chatbots begin by offering users a list of what the bot can help them with, or giving them an example of what kinds of questions to ask. Not only does this let users know what the bot can deliver, it also gives them ideas on how to further engage with the Chatbot.
Once you decide how to instruct your customers on interacting with your Chatbot, you need to choose how exactly they will take that journey. Will they type out text to your Chatbot? Or will they click on pre-set buttons that will take them up the trunk and through the branches of the bot’s decision tree? What about a combination of both – with buttons as suggestions but including the option for typed text as well? (For example, the mental health Chatbot, Woebot, uses both text and buttons interchangeably to help users look for patterns in their mood, and come up with strategies to improve it over time.)
Another thing to consider is how you will present customers with information. Chatbots of brands like Fandango and Whole Foods use carousels to give customers a list of scrollable options with interest-peaking photos. These carousels offer customers additional selections that are easily available – making for short, but effective decision trees.
Once you have planned out your Chatbot, it’s time to map your content. For the simplest and most intuitive approach to content mapping, we recommend a process that is focused around building decision trees from existing customer interactions. This way, there is no secret to Chatbot journey mapping. You already have most of the content you need to work with – you just have to put it in place using the following steps.
Self-service has been growing in popularity, which means that chances are, if you have a website, then you have a customer-facing knowledge base or an FAQ section.
Recommended for you: What is a Knowledge Base and Why is it Useful?
Customer-facing knowledge bases are used to help customers find the answers to their own questions. They are generally compiled of frequently asked questions – the same kinds of questions that customers who don’t feel like searching through your knowledge base may write to a live chat agent about.
Some knowledge bases are already presented in question and answer format. If that is the case, then you have some great starting ideas (and answers) for your Chatbot. If that is not the case, use your knowledge base as inspiration to create questions.
Customer-facing knowledge bases aren’t the only resources that companies have. Often contact centers will have their own agent-facing knowledge base. This kind of knowledge base helps agents respond quickly and accurately to customer queries. Agent-facing knowledge bases might be compiled as document and shared with all support agents.
Another resource – and perhaps the most valuable resource that you can consult – is your agents’ list of canned messages. Like agent-facing knowledge bases, canned messages are usually available as a shared document, or within the messaging system. Up-to-date canned messages can give you insight into the support questions that are most commonly asked of your agents – and exactly what the response to those questions are.
By consulting your knowledge bases, FAQ resources, and canned messages, you can begin to compile a list of what easily answered questions you can assign to your Chatbot.
For deeper insight into the questions that your customers are asking your live chat agents, assess your existing chat logs. This will help you discover the most recent common questions, and any questions that you might have missed out on when examining your knowledge bases.
Once you have identified the questions that your Chatbot is capable of answering, you can begin the conversation mapping process.
If you plan on offering a Chatbot that gives customers selection options, such as clickable buttons rather than text, then a good way for you to start conversation mapping is from the “trunk” of the tree, and work your way up towards the leaves. That means starting by sorting out your eligible questions into basic categories, and working your way towards the detailed “leaves”.
Divide questions into the broadest categories first, then into smaller categories. For example, let’s imagine that you are a clothing retailer. You can start by dividing your questions into two categories: Sales and Support.
In the “Support” category, you can then branch out into other options, such as:
The “Sales” category can have options such as:
From there, continue the tree, branching out further with each option. For example, if we continue to work with “Shipping and Delivery,” this category could branch out into the following questions:
Even if you plan on ditching buttons for the most part, and having your customers type their queries directly into the chat box, this is a great starting framework to help get you organized.
Work towards the leaves by taking your common queries and providing answers.
You can either offer simple knowledge base responses, or you can go the extra mile by personalizing them even further. For example, you can answer this question several ways:
Question: Can you ship to my country?
Option number one is an easier way of doing the same thing as the other two questions. However, what is more work for you a single time will be less work for your customers in the long run… and they’ll appreciate you for it.
Option number two abandons the button model and assumes that the user can type in an answer. Question number two also assumes that the Chatbot does not have access to the user’s location, and therefore cannot respond like number three.
Option number three is the most user-friendly of these options. It can be used with buttons or typed responses, and it assumes that the bot has access to the user’s location. However, should the “no” option be selected, the user must be able to type their country in.
And you guessed it – to use options two and three, you would need to go through all the countries that you ship to and all the countries that you don’t, and create branches for each of those responses. It’s a bit of work but hey, your customers are worth it!
Once you’ve answered your customer’s current question, you may be ready to move on but they might not be. Customers need a way to wrap up the conversation, or move on to another topic.
Let’s go back to the previous example, using the third option – only this time, let’s see what that looks like when you give your user a follow-up action:
Question: Can you ship to my country?
In this scenario, these options would be visible as buttons. If you let users type back to you, you will have to expand the transition options that are available, as users might jump from one thing to something unrelated.
Just because your Chatbot is artificial intelligence doesn’t mean that it has to sound robotic. Modify your responses to make them warmer and more human-like. Check out how we did this to the above example:
Question: Can you ship to my country?
Friendly representatives can ease tension, and make customers smile. Friendly, human-like Chatbots can do the same. By adapting content to reflect the norms of conversation flow, your Chatbot will be a lot more charming, and more pleasant to interact with.
Optimizing conversation flow isn’t only about making the conversation more natural – it’s about removing obstacles for the customer. For example, for questions that require more in-depth help or heavy information, you can share an informative video, photo, or a link to a knowledge base article, instead of having your Chatbot send the user a block of text.
Not all content has to be entirely related to solving the user’s question. Part of the reason that customers enjoy Chatbots is that they are fun to mess with. If engagement is one of your Chatbot’s goals, it doesn’t hurt to drop in a few treats for your users that will inevitably troll – or try and get closer to – your bot.
Come up with playful, human-like, or creative answers to comments to keep your users entertained after they have resolved their issue. Map out possible responses to questions like:
You can also offer creative content to your users as a part of your decision tree. Check out how Woebot does this below:
Who can resist that face? Even if your user didn’t get what they were looking for out of the bot, this creative, sharable content is bound to make them smile.
If your customers are typing answers to your bot instead of selecting buttons, you are probably going to need a lot of extra branches for your decision tree. You will have to account for the different ways that someone might ask the same question, in order to achieve a satisfying performance across the board.
Here’s an example of how the weather prediction Chatbot, Poncho did this wrong:
Source: Why Chatbots Fail
Here, the Chatbot failed to recognize the word “weekend,” a major faux pas for a weather prediction bot. The bot also didn’t account for normal human messaging patterns, which often involve follow-up questions, and repeating only the bit of the question that was missed.
Poncho was wrong about not reacting to this key word, but Poncho doesn’t have it all wrong. Let’s take a look at something that this bot did right:
Here, the Chatbot’s decision tree responded to the incorrect spelling of the word “Fahrenheit” as if it had been spelled right. By accounting for multiple spellings of the same word and misspellings, your bot will be able to help everyone regardless of their written abilities or typos.
You want your bot to do everything and know everything, but there will inevitably be some things that it just can’t do, or that it doesn’t understand. Miscommunications happen, and it’s important to be prepared for them.
“I’m sorry, I didn’t quite get that?” or, “I don’t know what you mean,” are good ways to let your user know that they need to repeat themselves, or try asking a question a different way.
To take attention away from what the Chatbot doesn’t know, emphasize what it does know. You can offer a list of suggestions to users. Here’s an example of what this might look like:
“I want to change my order.”
Another way to respond to something that your bot can’t do is to remind your user: your bot is just a Chatbot. Here’s an example of how you can use that to your advantage:
Eighty-six percent of customers believe that they should always have the option to transfer to a live chat agent when dealing with a Chatbot. While chatting on your website, be sure that one of your Chatbot’s responses for what it doesn’t know gives customers the option to speak to a live chat agent. This might look something like this:
“I ordered the wrong thing. I want to change it.”
Giving users a way to disconnect with the Chatbot when it is unable to help them and get support from a live agent makes sure that your Chatbot remains a tool for your customers, not an obstacle.
When introducing a new technology to your customer service strategy, there are likely to be some speedbumps. There will almost certainly be branches on your Chatbot’s decision tree that are faulty, and that need tweaking. There may be easily-answered questions that you forgot to account for, or elements of your bot’s personality that rub your customers the wrong way.
The important thing is that you work out the kinks in your Chatbot as you go along. With time, your Chatbot will only get smarter and more capable of assisting customers. By going over your bot’s chat log and listening to customer feedback, you can make changes to your Chatbot and gain experience that you can only really have through trial and error.
If Poncho learned – you can too! The weather Chatbot now has an expert response for the follow-up question, “This weekend?”
Here are some final tips to keep in mind when creating your Chatbot’s decision trees.
A Chatbot is only as good as its decision tree. Effective journey mapping can make any Chatbot – regardless of its level – a useful customer engagement tool and self-service resource.
We hope that this blog post has had a hand in showing you the power of decision trees. Keep these steps and tips in mind when it comes time to map the content for your own Chatbot, and buckle up for Chatbot success.
For more information about Chatbots, check out our blog post, It’s All About the $$$ – How much Money Can Chatbots Actually Save?