When you think of chatbot marketing, what comes to mind?
Opinions run the gamut from fear — “What’ll it be like to entrust my customer service to a computer?” — to humor — “Here comes the robot apocalypse!” But reality is that there are marketing teams and support teams and sales teams making serious progress with their chatbot strategies.
We want to help you be one of those brands with a rockin’ chatbot strategy.
In this post, we’ll go deep into the world of messenger bots to give you the details on how to use chatbots for small business and build a best-in-class chatbot marketing strategy. We’ll answer your questions about best practices for a nearly-human chatbot experience as well as how to get the most value out of chatbots on Facebook Messenger, Twitter, WhatsApp, and more.
Here we go!
First things first …
What is a chatbot?
A bot is simply a computer program that automates certain tasks.
In the case of a chatbot, the task being automated is a 1:1 conversation with a person.
These bots can use sophisticated technology like artificial intelligence and natural-language processing. Or they can be as simple as a series of IF-THEN statements. Alexa is a type of chatbot. So is the Domino’s Pizza app.
Today’s chatbots reply with text, yes, and also with audio, video, images, GIFs, you name it. Even the mediums for chatbots have grown exponentially — you’ve likely experienced bots in chat apps like Messenger and WhatsApp as well as on many, many websites with the little button in the corner asking if you need any help.
We’ll be focusing specifically on chatbots on social media channels in this post.
How to build your own messenger chatbot?
The good news: You don’t have to have a lick of engineering skills.
There are a host of services that allow you to build your own chatbot, no engineering background required. A lot of these tools are as easy to set up as figuring out what you want your IF-THEN sequence to be:
Facebook Messenger’s official page offers to build your own bot directly through the platform’s landing page. This method though, may be a little bit more complicated than others.
10 Tips for Your Chatbot Marketing Strategy
We ended up researching a number of different world-class chatbot experiences to learn from. Here is a list of 10 lessons for anyone about to get into chatbot marketing — like us.
Research your most frequently asked questions, by asking your team
One of the first things to consider with your bot is the content that it’ll contain.
Let’s take one of our favorite chatbot use cases as an example: a customer service bot. If the aim of the bot is to help customers and deliver speedy responses, then we suggest looking at the most frequently asked questions of your brand to see what content makes sense to start with.
To find these FAQs, there are a couple great places to look:
- Your customer service team. They likely will have a heap of questions off the top of their head that they hear from customers all day.
- Your social media support team, your community team, or your social media manager. Whoever does the engagement on your social profiles should have a good handle of common questions that come in through @mentions and DMs.
- Your sales and marketing teams will have a pulse on what types of questions they see as customers progress through the funnel. This could look like common Sales questions that reps face, or it could be questions that your content marketers are seeking to answer.
And of course you could source questions from outside of your immediate team, too. The search suggestions at the bottom of relevant Google pages are a good place to start, as are crowdsourced communities like Quora and Reddit.
If you choose to build a bot for sales, lead gen, or any other service, we highly encourage you to research common questions and customer journeys so your bot is fully prepared to be as useful as possible.
Build your bot its very own conversation tree
Chatbots work best when given a concrete set of questions to answer. Without a certain level of specificity and pre-planning, then it becomes infinitely harder for a chatbot to deliver a believable experience — much less the right answer.
This is why a conversation tree works so well.
Picture a gigantic flowchart or a mindmap. Beginning with the initial hello from the bot and its very first ask of the user, you branch off from there, building the conversation flows for every different direction the conversation may turn.
We’re big fans of tools like Lucidcharts and Whimsical for creating easy-to-read flowcharts that would suit this type of project perfectly.
Avoid fully open-ended conversations
Open-ended conversations can lead to confusion for your bot and a poor experience for the user. If you don’t have the luxury of highly-advanced language processing, then an open-ended question like “how can we help you today” could go any number of directions.
One of our favorite chatbots is the one at Hello Fresh in Facebook Messenger. Among the bot’s first messages to the user is an offering of a menu of choices: “Here are some common questions I can answer” Options include things like:
- How does it work?
- What does it cost?
- Are you gluten-free?
- Are you vegan?
- Give me a discount!
The user can choose any of these statements by tapping on them in the Messenger interface. Then the bot will respond with an automated reply.
This takes the guesswork out of the bot’s replies since it knows exactly what to say to exactly which message it receives.
Let people know that a human is just a step away
One of the most interesting stats we’ve seen about chatbots is that people aren’t nearly as turned off by them as you’d think. 69% of consumers prefer communicating with chatbots versus in-app support. People love speedy answers to their problems.
That being said, that leaves 31% of consumers who might prefer the old-fashioned way — email or social support.BROUGHT TO YOU BY
Never miss an update from us. Join 100,000+ marketers and leaders.
This can be baked into your bot experience, too.
Simply let people know as part of the bot’s welcome messages that the user is invited to get in touch with a human at anytime.
Give your bot a voice … and a warm welcome message
Just like you do with the way you write as your brand on social media, you’ll want to think about the voice and tone of your chatbot as well. Perhaps this is simply a natural extension of your brand’s voice and tone.
Other companies choose to lean into the “bot-ness” by making the voice a bit more obviously robotic.
Whatever you choose is entirely up to you. Just stay consistent with it throughout your conversation tree.
And one of the most important places to nail this voice and tone is in the opening message from your bot. We mentioned in the previous tip to be sure you let users know they can get in touch with a human at anytime. That’s a great nugget to place into your bot’s welcome message.
Also look to include things like:
- A catchy hello.
- Expectation-setting. Letting people know they’re talking with a bot.
- And a solid first question with plenty of options to capture as many possible user journeys as you can.
Track the effectiveness of your bot with special UTMs and discount codes
One of the biggest questions you probably will have with your chatbot is … is this thing working?
And “working” can mean a lot of different things. If you’re using chatbots to minimize your customer support volume, then that’s an easy metric to check. If you’re wanting to measure the effectiveness of education, marketing, or sales, then it can be invaluable to track the bot’s success with measurable links and codes.
Hello Fresh does this with their bot — including the word “bot” right in their coupon codes (for example: FRESHBOT25).
Similarly, you can do this with your UTM codes for the content you link from your bot. Give it a UTM source of chatbot and you can measure the clicks and traffic that come from the bot, as well as track the UTM all the way through your customer journey.
You may even end up measuring ROI from the bot, which would be incredible!
Replace your email newsletters with chatbot newsletters
We’ve talked a lot about how great a chatbot can be for incoming requests. But how about outbound? There is a lot of room to experiment here. And one of the prime places is using your bot as a content delivery system.
For instance, on Facebook Messenger, any time someone talks to you through Messenger, they are added to your contact list. You can set up a chat bot to ask these folks to opt in to hear regular updates and announcements from you, then — voila! — you’ve just built a subscriber list on your messenger bot.
Tools like Mobile Monkey can then make it easy to send out new blog posts or quick information to this group. Some estimates say that chatbot newsletters generate a 70-80% engagement rate.
Send simple surveys to your contacts
With the high engagement rate with bots, you also have a good chance of getting your message noticed for surveys. It can be notoriously hard when surveying folks via email or on a website or app to get a high volume of responses. It’s a bit easier with bots.
Similar to the email newsletter tip above, with surveys, you first ask people to opt in to hear from you, then you can message them occasionally with a short and simple survey.
Many of the tools we mentioned earlier include the option for two button-based responses, which are perfectly suited for the mobile-first experiences of social media bots.
Enrich your bot with data and personalization
Check out this list of powerful chatbot superpowers:
- Universal Studios tells you the wait time for rides
- Marriott can tell you room availability
- CheapFlights tells you the best options for your dates and your price range
- Domino’s lets you order pizza!
These are all possible because of the Big Data that these brands pipe into their bots. If you’re not quite at this scale yet, no worries. You can dip your toe in the water by anticipating the most common questions of your customers and doing your best to fill in your bot with details. Simple things like hours of operations, daily deals, etc. can make for a delightful experience.
And if you do have a customer base who clamors for data-rich answers, then use the examples above to inspire your chatbot dreams.
There’s also the matter of personalization. And for this one, we’ll leave it up to your best judgment. Many tools allow you to personalize the chat experience with variables like first names or locations. This tows the line between helpful and offputting, when coming from a bot. Use discretion.
Make sure to promote your chatbot so people know you have one
This one might seem obvious, but it’s perhaps one of the most important tips we’ve covered so far. Your bot will only be successful if people find it and use it. So get the word out!
This can happen organically as people visit your Facebook page and are routed to you on Messenger. Or you can be proactive about it.
A couple of our favorite ways of promotion are:
- Mentioning your bot in a list of all your customer support channels
- Adding a call-to-action on your blog or website to get in touch with you
- Advertising on social for people to opt-in to your bot experience. This can be great for the Messenger newsletters we talked about.
How to say hello to us
We would all love to say hello to you on social media – especially Twitter!
Thanks for listening! Feel free to connect with our team at Buffer on Twitter, Buffer on Facebook, our Podcast homepage, or with the hashtag #bufferpodcast.
Enjoy the show? It’d mean the world to us if you’d be up for giving us a rating and review on iTunes!
About The Science of Social Media podcast
The Science of Social Media is your weekly sandbox for social media stories, insights, experimentation, and inspiration. Every Monday (and sometimes more) we share the most cutting-edge social media marketing tactics from brands and influencers in every industry. If you’re a social media team of one, business owner, marketer, or someone simply interested in social media marketing, you’re sure to find something useful in each and every episode. It’s our hope that you’ll join our 27,000+ weekly iTunes listeners and rock your social media channels as a result!