Answer These 5 Basic Questions Before You Build A Chatbot

Here a bot, there a bot, everywhere a chatbot—there’s no denying that chatbots are drawing a great deal of attention lately! You may be asking yourself, “Is now the right time for our business to hop on the chatbot bandwagon?”

If so, you’re not alone: Many companies are exploring the benefits of using bot technologies to support all kinds of business use cases. Just take a look at these 50 examples of how brands are using chatbots. 

Chatbot technology represents enormous potential for enhancing the customer journey. You can gain a huge competitive advantage by allowing customers to interact with your brand through a conversational interface. They simply ask a question, and the bots figure out what they need and how to make it happen. Instead of requiring customers to do all the heavy lifting, brands can meet them where they already are and communicate with them on their terms.

As with any initiative, you should establish the scope of your chatbot implementation before getting started. Here are the five key questions that are vitally important to answer before launching a chatbot:

5 Basic Questions You Need To Answer Before Building A Chatbot

1. What will the chatbot do?

Many brands have launched chatbots to allow customers to access information via messaging apps and virtual assistants. Examples include asking your Amazon Echo what your checking account balance is, or receiving a Facebook Message when a package has been delivered to your home. With both of these examples, chatbots simply allow customers to interact with information that’s already easily available through a website or mobile app.

But this emerging technology is rapidly becoming more sophisticated, developing the ability to engage with consumers to accomplish more complex tasks. Chatbots can help customers perform product research by asking questions to narrow down options (much like a store associate would do). If the customer is ready to purchase, chatbots can handle that, too—accepting payments, sending shipment alerts, and answering any questions that come up—all within the same interface the customer is already using. Check out some surprising examples of how chatbots can enhance the customer journey.

Ultimately, what you want your chatbot to do will depend upon which part of the customer journey you most want to improve. Start by mapping your current customer journeys and identifying places where chatbots could make the biggest impact.

According to Microsoft, nine out of 10 consumers expect every company to offer 24/7 online self-service. With that statistic in mind, you may want your first chatbot to answer customer questions only when your contact center staff is unavailable. 

Eventually you might want your chatbots to be able to do everything your customer service agents do. Agents free up to handle challenging and sensitive escalations and focus on adding more value to each interaction by collecting voice of the customer data, up-selling and cross-selling, educating customers in order to head off future issues, and building emotional connections.

Whatever your aspirations, approach your first chatbot project with a clearly defined set of goals and an understanding that you’ll have to walk before you can run.

2. Where will the chatbot live?

To put this another way, where will people interact with your chatbot? Here are some channels to consider:

• Your website
• Your mobile app
• Facebook Messenger
• WeChat
• Kik
• iMessage

To ensure the highest adoption rate, you should start with the channels that are already being used by your customers. Forrester found that one in every nine people on the planet use Facebook Messenger in any given month. But unless your target market is everyone on the planet, you should delve into which channels are most popular with your current and prospective customers. If you sell primarily to users of Android devices, iMessage might not be the best platform to start with.

3. What information sources will the chatbot use?

No chatbot can do its job without access to data, whether from your website, knowledge base, existing documents, reservation systems, shipping information, product inventories, partner sites, or some other source.

As part of your chatbot strategy, you’ll need to decide what sources of information to make available to the bot. This decision depends a great deal on your answer to the first question (What do you want the chatbot to do?). Say you’re an auto brand that wants a chatbot to help customers schedule service appointments at their local dealer. To perform all the tasks associated with this transaction, the bot will at least need access to:

• Information about the car’s make and model
• Service history
• Account information (i.e., if the customer has a service package)
• The dealer’s scheduling and billing systems

After you have determined which data sources will be needed to perform each task, you can map out how the bot will access each source. Will it be able to pull an answer from your existing knowledge base or website? Will you need to build a custom integration, or is there an open API you can use? Will customers need to enter any kind of PIN or password to allow the bot to access certain data, such as sensitive financial or personal information? Can your business systems currently communicate with each other, or will some extra legwork be required?

Mapping out the details of your future chatbot’s data sources may sound like a tedious task, but you will thank yourself later. 

4. When will the chatbot escalate to a human?

Even the most sophisticated bot will eventually encounter a problem it cannot solve. That’s why it is crucial to determine those limits, and decide what will happen when your chatbot reaches one of them.

An escalation to a human agent might be triggered when a customer:

• Requests a phone or video call
• Uses language that indicates frustration or anger
• Asks the same question multiple times
• References a potentially sensitive or dangerous issue, such as an allergic reaction or a medical emergency
• Asks a question the bot doesn’t know how to answer

In some of the above examples, the bot is called upon to infer an aspect of the customer’s situation or state of mind based on something they have said. Sentiment analysis, natural language processing (NLP), and machine learning play an important role here. For instance, the customer may describe their symptoms instead of saying the exact words, “I’m having an allergic reaction.” The bot must be smart enough to recognize what’s happening and escalate the customer to a human.

And when the hand-off happens, brands must ensure a seamless transition. Best practices include:

• Allowing the customer to continue the conversation using the same channel
• Avoiding asking the customer to repeat their story (since two-thirds of consumers say they are extremely frustrated by having to repeat information across multiple channels, according to data from Accenture) 
• Enabling the customer service rep to pick up where the bot left off by providing them with relevant context and interaction history

Another aspect of escalation to consider: how obvious will it be to your customers when they are talking to a bot vs. a human? Depending on your objectives, you may want to explicitly tell customers they are communicating with a bot; for instance, if you’re promoting a new animated movie, you could invite people to chat with one of the characters. But if, for example, you’re helping a customer set up Bluetooth in their new car, you may want the entire interaction to feel the same as a conversation with a human.

5. Who will build the chatbot?

Although some brands are leveraging internal resources to build a homegrown bot, most are teaming up with external vendors to get their chatbots up and running. 

As you research potential vendors, look for some of the following attributes:

• Broad capabilities for enhancing the customer journey
• Robust customization options
• Expertise in your industry
• Ability to implement across multiple channels and to use multiple data sources
• Strong NLP, machine learning, and sentiment analysis capabilities

One final consideration: pick a company that can not only help you get your first chatbot iteration off the ground, but also has the vision to help you develop the second, third, or even 20th! We haven’t even begun to scratch the surface of what chatbots can do, so it’s wise to choose a vendor that can grow with your brand and with the technology.

Before you take the leap into the exciting world of chatbots, keep in mind 1) what it needs to do, 2) where it will live, 3) what information it needs, 4) when it will escalate, and 5) who will build it. Answering these five questions will set you—and your chatbot—on the path to success.