Heading into 2018, you may have started to notice a lot of buzz around the word “chatbot.” What exactly is it? How does it work, and is it for everyone?
The subject of chatbots has become more prevalent since Facebook Ads unveiled the option to have ads drive users to Messenger. It helped reduce the steps and friction around driving users to a business website, where business owners were hoping for a site chat engagement or a contact form fill. Messenger also feels very personal in nature, but scaling it without a larger social team can quickly be a non-starter for smaller companies who have customers reaching out via social.
Chatbots offer a way forward for advertisers who want to provide customer service in a semi-personal way, without having a "single person" handling all the volume. They act as a pre-programmed script, so users can get helpful or needed information in a way that feels tailored to them.
Examples of Chatbot Uses
Chatbot platforms are becoming really versatile, really fast. Here are a couple of examples:
I know, I know. “Those are humongous brands, how does that help me?” That is the beauty of chatbots with Messenger! They can be used for so many things, no matter the size of your company.
Chatbot platforms essentially help you create a flow of conversation. You can have a lot of flows, which is perfect because users usually have a lot of different needs. One of the things I love most about this emerging tactic is the experience is really up to you:
- They can be a fantastic way to supplement email marketing efforts, allowing you to broadcast updates and content to subscribers.
- They can operate as a customer service platform to help answer questions, and get someone in touch with a live person if they need that kind of help.
- It is yet another channel to distribute video content.
The possibilities go on and on.
How are Chatbots Made?
The platforms all look a little different in their execution and user interface, but they follow a basic equation.
You create a chat flow first. This will vary depending on who you are talking to and the context in which they are contacting you.
For example, maybe the bot sits on your website and invites people to subscribe to updates anytime you update your blog.
Maybe the bot sits in a remarketing ad on Facebook, inviting users who abandoned their cart to reach out if they have any questions you can help with.
Whatever the situation, there are several basic functions in the platforms:
Flow: This is what it sounds like - a guided conversation, where you ask questions, users answers, and based on that answer they receive new choices. It is a lot like the old Choose Your Own Adventure books.
Triggers: These are based on things like keywords a user types in. It literally triggers a conversation flow. For example, if a user uses the word “help,” your bot will put them into a conversation flow to help them.
Promotion: These are the options available to help get bot subscribers. Generally, they are things like JSON code that can go into a Facebook ad or a site modal that prompts users to sign up. I have a modal on my site for people that scroll to a certain length, for example.
These sound like simple buckets, but it gets complex fast when you realize how man possible conversations you can have with customers!
The chat bot platforms guide you through creation with a visual interface. Here is an example from MobileMonkey’s Builder:
MobileMonkey text widget input field
This has a pretend welcome message for people who want to subscribe and get a free guide on how to build a chatbot. I input the message I want the user to see when Messenger opens for them. You will see there is an option to add a button; this is where I can add buttons a user can hit, and based on what they choose, they go down the conversation path:
MobileMonkey button editing
So, you enter the button text a user will see, and what you want it to do…do you want it to open another window with more questions? Dial a number if they hit it? Go to a certain URL?
This is how the conversations are built. Module by module, you build out the potential conversation.
My Personal Case Study
I initially launched a bot with ManyChat on my site, and then also tested it in a retargeting ad on Facebook. Instead of a typical Facebook ad that drove traffic to my website, it dropped the user into Messenger where my bot would kick off the conversation. I was already running remarketing ads that drove people to my site to get new subscribers on my email list and was curious to see how the bot's performance would compare.
I created a welcome message in ManyChat, gave information for after they signed up, and also offered to direct them towards more information Facebook Ads. They could decide if they wanted information from me (my blog) or recommendations from other people.
Here is what the welcome flow looks like when you view it in ManyChat’s visual flow:
The chat results vs. site-driving to email signup were really interesting.
Of the users who received the chatbot-focused ad and clicked to sign up for updates, 100% of the users opened the welcome message. 100% of them interacted with the bot. (I sort of equate bot interaction with clickthrough rate on an email - I know, it is not apples-to-apples.)
Ultimately, the money I tossed towards retargeting ads for my bot resulted in a $2.13 per sign up.
Now let's look at the ads that focused on driving users to a landing page to join my email list. Those cost $8.01 per email sign up and had a 23% open rate. Clickthrough averaged 27%. So, early on, the bot is winning for cost to acquire and user engagement vs. the traditional method.
What made this even MORE interesting is the targeting for the email sign-ups were people who had been to my sessions at conferences. I had a personal connection with them, they had downloaded a worksheet I made available, but their engagement was LOWER than users who had just visited my site.
Chatbot Curiosity is Partially the Key
It is amazing for me to see the results with clients in their unique testing situations; they are not dissimilar from mine. Chatbots are not widely used, so for users, the newness leads to a curiosity factor that leads to testing on their part.
I think in the long-term, as the curiosity part wears off, the engagement with them will remain strong. They are more personalized than email because they are interactive. They feel more real-time than an email, and users can reply, interact, and guide the bot towards what they care about. The chatbot platforms also have the advantage that a human can jump into the flow at any time to take over. It helps manage and scale the flow of communication in a way that feels more human, with the option of actually being human 100% of the time.
Clients who were a little scared to venture into the chatbot realm at first are getting really excited. Most started off basically creating their FAQ as a chatbot flow and were understandably nervous. However, as they see more conversations happen and the valuable information that can be mined from them, they get really gung-ho about launching more flows.
Learnings, and Building for the Future
All of the chatbot flows have a record, so you can go back through and see stats on the buttons users are clicking. It gives tremendous insight into what they care about, and where you might want to create additional segments.
This data also allows you to keep making your flows smarter. You can scan for common keywords such as “help” or “hours” or any number of things, and create a new conversation flow around them to service those requests — simply by the keyword being typed by a user.
Your bot has the ability to become a living customer service machine, and an active promoter for your content and brand. While it takes a little effort and mapping, it saves so much time to answer those same questions without using your staff to do it or relying 100% on flooded email boxes.