Facebook Messenger Chatbot Storyteller

I grew up with Choose Your Own Adventure (CYOA) books. Now in the era of chatbots, can I harness that same magic?

Problem

Marketing is hard. Email open rates are low. Online advertising is being attacked by Apple and Google (good for consumers, bad for small businesses). People want instant gratification and free stuff.

So, how do I stand out in a crowded market using marketing channels people are using while providing the instant gratification customers are demanding while not spending a fortune in time or money?

Project: Breadcrumbs Continued

Growing up with CYOA books and Dungeons & Dragons, I have a love for interactive stories. At my work I've been helping companies build digital assistants to complete tasks like processing invoices, checking data for errors, copying data from application one to application two, as well as customer service chatbots. This last one made me stop and think, what can I do with that?

Stories are interesting and communicate all kinds of things but audience involvement makes them deeply engaging. Chatbots are conversations between a human and a computer. Could the computer be the storyteller if I taught it how to tell the story?

But, chatbots aren’t cheap. Spinning up a chatbot system can get expensive and take a lot of time, testing and tears to get it working how I’d want it to work. Then comes the challenges with deploying to my website and making it available on platforms that people can easily use…yuck.

Instead, I tried another direction. I built a chatbot for my Facebook Author Page using Facebook Messenger. This way anyone who has a Facebook account, including Instagram, Oculus, WhatsApp, could access the chatbot and I could tie it into Facebook’s marketing capabilities.

Using a visual design tool, called ManyChat, I prototyped the conversation flow and story that I wanted to be told. Adding interactive elements were easy and important to chop up the large blocks of text.

Using ManyChat I was able to deploy the chatbot to Messenger with options (not tested yet) to also deploy to Instagram. After testing the bot, I added an audio component to make the story also available through audio so people who didn’t want long text reads could just listen.

Check out the final product here.

Learnings

Plan the conversation AND the story

I originally put a lot of focus on the story but once I started laying out the conversation, I had to rework various elements. The content length requirements, the engagement, the drive to see what’s next all had to be right.

Do not approach this process as write your story in word then copy and paste into ManyChat (side note: I didn’t approach it like that but learned a lot I’d do differently in the story drafting process).

The conversation is key to the storytelling experience. Help the listener stay engaged through prompts, questions, choices. Imagine if you were telling this story to someone else, you want to give them the moments to ask “and then what happened?” Similarly, when you are designing your conversation, give the human an active role beyond “click next to continue”.

Not just talking

The first iteration of the chatbot had GIANT text blocks that my testers just did not read. Don’t do that. Break up your text into smaller chunks and use other media to tell your story. Messenger allows you to share pictures, audio, video, and other interactions to engage your audience. Think about your story across all these lines.

Could you show a picture of the farm your story takes place in? Is there a video that goes with the section of the story happening right now?

One of the features I used was adding Audio to the bottom of the story sections. The audio was a reading of the message contents. This allowed the reader to turn the conversation into an audio experience which made the story more engaging (so the testers said).

Decisions, Tags and Flows

Your story isn’t just a bunch of text blocks. It is an interactive experience. To that end, you need to track if people are using it, if they are leaving at certain places, what they are doing and what kind of opportunities that opens for you.

This isn’t as complicated as it sounds. Using tools like ManyChat allow you to visualize the conversation flow as well as create “checkpoints” to tag a user. I used tags to know if someone had reviewed my book BLOTS or not by tagging the user after I asked the following question: “Have you reviewed BLOTS on Amazon?” If they say “Yes” then I tag them “Reviewed BLOTS”, if they say “No” I don’t tag them. This way in the future, I can create specific experiences for people who have reviewed BLOTS. Possibly an exclusive experience that is only available to those who have reviewed the book.

Using the tags allow you to personalize the experience.

Identify your goal, build to it

Why are you spending your time with a chatbot? For me, it was about creating a different kind of storytelling experience. As I dug into it, the project became more of a storytelling exercise that enabled marketing for the future. Now, I know who has had the first conversation with the main character and have asked what kind of stories people would like to see in the future. The guidance from readers is market research for future stories.

In exploring what could be done, I found a better picture of what I wanted to do. While I suggest having a goal in mind, don’t lose site of the value of wandering. Who knows what you’ll learn or where you’ll end up when you give yourself a chance to explore the unknown.

What’s next?

In my current implementation there is a lot of cleanup needed. The tagging is rough and could use optimization as well as improving the engagement of the text blocks to keep the reader going (I know that from the metrics).

Beyond Facebook Messenger, I want to explore Instagram and the integration with Oculus in Virtual Reality. Conversational interfaces have huge potential in VR and who knows, maybe one day we’ll be cozying up to a VR campfire with an AI storyteller talking the digital night away.

Cover Photo by Miguel Á. Padriñán from Pexels