Three Lessons from Three Chatbots

It’s easy to get excited by Artificial Intelligence. After all, barely a decade ago we were still getting to grips with these things called ‘apps'...
Let alone delivering ourselves home by driverless car; bathed in the soft lights and sounds of automated Hue routines and algorithmically curated Spotify playlists.
However, while this scene is slowly becoming a reality for some, we should remember that the expense of kitting yourself out with the necessary gadgets and broadband speed makes full automation a way off for most.
Meanwhile what some 85%* of us do already have is our smartphones. And after mastering the app game some years ago, it’s here the likes of Siri, Google or one of a rapidly expanding catalogue of Chatbots on Facebook Messenger are offering people the opportunity to experience AI first hand.

So what is a Chatbot?

At its most simple, a Chatbot is a product or service that people interact with in a conversational way, either by typing or using their voice.
Exactly what the product or service is can vary wildly, from booking a hotel to playing a game, and from self-contained applications to fully-integrated machines that connect with other data sources and processes. The only constant is conversation.

Sounds exciting, huh?

We thought so, so we went ahead and built three. Here’s what we learnt:

Lesson One: Understand Your Goals

Building a Chatbot for your brand or business is no different to building any other digital product or service. If you’re going to get the go-ahead for the project, you need to have a goal in mind.
What do you want the Chatbot to achieve? What problems might a conversational interface solve for your customers? How are you going to measure your success?
The key is to start the design process armed with answers to these questions and build backward from there.
When we built the Inc.redible Gift Bot for our client, Nails Inc, we knew our solution needed to turn heads and drive sales during the busy Christmas period. These goals helped direct our approach, incorporating trackable links and exclusive bundles throughout the Chatbot’s decision-tree logic, helping us track and optimise the user journey toward the products attracting the most attention. The result: a record-level conversion rate and a shortlist at the Campaign Tech Awards.

Lesson Two: Conversations Aren’t Linear

It’s easy to mistake the way we talk with friends and family as a coherent narrative. I ask a question, you answer, I respond and we go from there. In reality however things are a lot more chaotic, leaning on shared reference points and regularly digressing on various tangents (pun intended).
If, then, we design our Chatbot as a linear process from A to B, request to response, the whole experience quickly breaks down and we lose the unique benefits of conversation.
Instead, successful Chatbots allow users to jump around, linking back through easily-accessible menus and remembering what people have seen and said before.
When building a Chatbot for the recent Customer Experience Conference here in London, we invited delegates to learn about the speakers, see the schedule, and ask questions through a comprehensive on-boarding process. By assigning attributes to users at the point they accessed information, we avoided repetition, and redirected them to other areas of interest.

Lesson Three: Personality is Key

Yes, it’s important to let your users know up front that they’re talking with a machine, not a human. But that’s not to say a Chatbot should be void of human-like personality. After all, half the point of embedding within apps like Facebook Messenger is to mimic the ways people already use them to talk with one another.
People’s expectations of platforms like Messenger or Whatsapp are personal, so if a brand or business is going to be welcomed into this private space, they need to offer a reciprocal sense of intimacy and openness you’d expect when talking to a friend.
Inspired by our love of Netflix’ Stranger Things, we built El Bot, a Chatbot that enables fans around the world to chat with the show’s ‘Upside Down’ world. By using custom GIFs and images to respond to users, rather than standard text-based answers, not only did we challenge people’s expectations of what a Chatbot can do, but also provided a fun, personal experience that fit Messenger’s content-hungry users.