How will you make it successful? What sort of tips and tricks can we, expert chatbot makers, give you?
This is the purpose of this article. I want to share a few tips (and some tricks) on how to get the most out of your project. A successful chatbot starts with you. You're on the right path, keep reading.
I divided this article into three defined sections.
Engagement & retention
Development: building for success
The right chatbot development sets you up for success. While you start planning your build and writing down a few lines of code, keep the tips below in mind.
Your chatbot's one true goal
This is the one and mighty tip for you during the development phase. Remember this one at every step of the way, through every decision, during every feature discussion.
Satisfy your chatbot's One True Goal.
We talk about this all the time but it's somehow not enough. Your chatbot must solve one unique problem. Its sole goal is to solve that problem better than anything else could.
Keep this in mind when:
Planning features: will feature X help your chatbot achieve its goal? If yes, how?
Building: is the way we are building this scalable? Are we helping the chatbot reach its one true goal faster?
The success of your chatbot highly depends on this. Remember the chatbot lawyer we presented a little while ago? Its entire success is due to the fact that it solved a unique problem in a unique way.
Build for scale and ubiquity
Chatbots are popping on all sorts of channels. You can now deploy them on your website, on Facebook Messenger, on Line, etc.
Each of these channels is a gateway to a potentially massive audience. You must keep one of the key learnings from our chatbot research in mind: consumers want chatbots for convenience.
Conveniency is important for chatbot users.
Don't trap yourself within one channel (especially earned medias like Facebook Messenger).
Build for ubiquity. Your customers might be on Facebook Messenger now, but what if all the sudden they move to Instagram? Will you rebuild it all? Have a plan for ubiquitous building, either with your chatbot agency or with your developers.
Then, build for scale.
You won't believe how quickly things get out of hand if you don't prepare. At ubisend, we don't just build ubiquitous chatbots (it's in our name, by the way, ubisend!) -- we also build chatbots that can handle the heat.
The simple reason for that is chatbots are taking off right now. If you have even a decent size brand, you will get a lot of traction through your chatbot. Make sure your bot and its servers can handle the load of conversations.
Launch: the big push
Ahhh, launching a new product. The perfect cocktail of excitement and fear.
I will split this section in two: company-facing chatbots and customer-facing chatbots.
Launching a company-facing chatbot
The biggest tip I can give you about launching a company-facing chatbot is education.
Tell your people what the chatbot is. In fact, this should actually happen during the development phase. Don't launch without taking the time to talk to everyone about the bot.
Tell them what its purpose is. Be present to answer any questions about it. If possible, have someone technical (or from the chatbot building agency) present.
I have sat meetings and always found them valuable -- for all parties involved.
Doing this right, you will ensure a successful launch of your chatbot. Fail to do this, you will encounter disgruntled employees who may go out of their way not to interact with your bot.
Launching a customer-facing chatbot
I get reminded of this time and time again, most recently (re)watching a great talk by Matthew Barby. In it, he reminds us all of the importance of timing when launching a new product.
I believe this applies to customer-facing chatbots too.
When possible, align your launch with something happening in the world. Something you can piggyback on.
Launching a chatbot that educates users about wildlife in Africa? You might align its launch with a new law around trophy hunting.
I can already hear you: 'what if nothing is happening? Do I hold on? Or give-up?'
Of course not. If nothing is going on, make it happen yourself. It's going to be a steeper curve and harder to achieve, but you can create your own buzz.
The point is don't launch at random. Try to build up to it. Much like employees need to be educated about the company-facing chatbots, your customers must be educated about your new chatbot.
Engagement & retention: success is sticky
Finally, here we are. You've planned it. You've built it. You've launched it.
Now is the time to grow it.
Success in a chatbot is measured in engagement and retention. A successful chatbot has users who engage with it and users who come back to it.
To improve engagement, I have two tips:
Number one: onboard your users properly. Tell them what they can do with your chatbot. Give a clear explanation of the things they can achieve and how to get there. The more freedom you give your users, the less engagement you will see.
Number two: get them to the result quickly. Remember, your customers want chatbots for their efficiency. They don't care about your intricate, super-mapped out user journey. They want to get to the point, now. I would even argue you should dial the conversational UX right down when dealing with a clear user enquiry. User asks, chatbot responds -- that's it.
Retention is more tricky. Depending on the type of chatbot you are releasing, retention may not even matter to you.
The one thing to keep in mind is that retention is user-driven. Try as you might, you won't get people to come back to your chatbot by pushing notifications in their faces 24/7.
Deliver a great user experience, and they will come back. Go back to the two points above. Tweak these and observe your cohorts. Are the more educated cohorts coming back more often? Do your users tend to stick around when you dial up the conversational UX, or when you dial it down?
Making a successful chatbot
It takes practice and tweaking. It takes knowledge and experience. It also takes a little bit of luck.
The biggest tip of all: keep iterating. If at first you don't succeed, try 1,927 more times. Then some more.