The SF AppWorks Blog

How To Think About Chatbots For Your Business | SF AppWorks

Written by Andrew Greenstein | Apr 11, 2017 7:59:00 AM

We’ve been busy talking about and building chatbots this year, so we wanted to share a bit of our thought process when thinking about how and when to integrate chatbots into the digital strategy of a business. 

 

There are several advantages to creating a chatbot. They open a direct line of communication to your users. They allow you to quickly test out new distribution channels like Facebook Messenger and SMS. They reduce the need for customer service resources. They create magical user experiences. However, for chatbots to be effective they need to be carefully crafted to address very specific use cases. The more focused the use case, the more natural the chatbot will feel. 

 

WHAT IS A CHATBOT?

 

Boiled down, a chatbot is really just a way of querying a database, typically with natural language. The input can be voice (Siri, Alexa), text (Facebook Messenger, SMS), point and click through a web widget, or really any manner of communication, so long as you have something to translate your voice/words/hand gestures into actions that a computer can understand. 

 

 
 
With the proliferation of voice driven search, chatbots can even exist with no visual interface, as is the case with Amazon Echo and Google Home. In fact, Gartner predicts that by 2020, 30% of all searches will be conducted without an interface.
 

What distinguishes chatbots from classic search queries is that chatbots will try to determine the intent of your words, usually through various natural language processing mechanisms. When a computer understands your intent, it can provide you with precisely the information or content you are seeking. 

 

START WITH THE USE CASE

 

The most efficient way to go about building a chatbot is to come up with a very limited, very specific use case. A chatbot that does one thing really well is better than one that does several things poorly. You can always add more functionality later.

 

Below are three examples of chatbot use cases from our customer conversations:

 

 

TYPES OF CHATBOTS

 

Say you sell shoes. People can browse your shoes on your website, order them, and have them delivered to their home. You also have a store where people can come in and browse your shoes. Now let’s say you want a chatbot. What should it do? Help users find the right kind of shoe? Help users navigate your store? Help users check the status of their order and schedule a return? All three situations are well-suited for chatbots, but by trying to do all three with one chatbot, you will end up with one bot that doesn’t quite go far enough with any user flow. But if you think of each case independently, there is quite a bit you can do and I’ll show you what I mean. 

 

The Personal Curator Chatbot

 

You have thousands of shoes, all organized by style, purpose, and size, but browsing through them is tedious and your users rarely get past the second page of search results. You decide to build a bot to help users find the perfect shoe. You ask them to describe what they are planning to do (run a marathon, do some urban hiking, go boating) and the bot determines the right shoe type and style for that very specific activity, significantly narrowing the results. Users are able to quickly find the perfect shoe from thousands of choices with just a few words, or even an image. As a result, you see higher conversions on your website. 

 

The In-Store Chatbot

 

Your brick-and-mortar location is huge, and you have limited staff on hand. Customers usually have to wait to ask questions about inventory or to request a shoe size that is not on the floor. You decide to create a chatbot to help with those two requests. Customers enter the store and there is a sign that says “Text 555-5555 to get help with inventory or sizes”. Customers pull out their phone and text, and a personal shopping bot responds with “what can I help you with?”. Customers can say things like “Where are your running shoes” or “Does this shoe come in a size 10?”. The best part is, customers don’t have to download anything. A week later, a text message is sent to the customer to notify them that they are having a spring sale on socks, which your store also sells. 

 

Or you could build this bot as an app that would also allow users to scan barcodes to check inventory, complete purchases on their phone, or request different sizes. Now that they have an app installed, you can send them push notifications when it is time to get new shoes, or offer them run tracking. Once you start down the rabbit hole of a specific use case, you can create hyper-curated user experiences that will give your customers a ‘wow’ moment that they will share with others.   

 

The Customer Service Chatbot

You decide to build a bot that cuts down on the hundreds of phone calls and emails your support staff handles each day. Users can chat with your bot on Facebook, enter their order number to track their delivery, and provide feedback. If a shoe doesn’t fit, they can request a new size be sent.  You can roll this out to some of your users and see if the number of customer support calls go down. If they do, you can consider adding other interfaces, such as a website widget or an Alexa Skill, to further explore new distribution channels.  

 

It takes some creativity and thought to come up with clever and engaging use cases for your customers, but if you take the time you will create exceptional customer experiences while reducing operational costs. Lucky for you, we love to brainstorm, so shoot us a line and tell us about your business and we’ll help you come up with some good stuff.