You can set up AI customer service to automate 80% of customer service questions that come through your Facebook Messenger Chatbot (or a chat popup on your store), using Google’s Dialogflow AI.
When you’re growing a store and a brand a lot of entrepreneurs just focus on marketing they get hyper focused on Facebook ads and conversion rates and things like that.
This is fine.
However the most successful e-commerce brands also put a huge emphasis on customer experience.
Jeff Bezos said If there’s one reason we have done better than any of our peers in the Internet space over the last six years it’s because we have focused like a laser on customer experience.
One of the ways that you can improve customer experience is by replying to messages from customers and potential customers faster.
This is quite hard when you’re doing everything else you need to do it.
A successful brand so in this video I’m going to show you how you can automate 80 percent of the work of replying to customer questions through Facebook Messenger.
So why Facebook Messenger.
Well you’re already running Facebook ads which encourage customers to comment and message you.
People prefer chat to email and Facebook Messenger is the biggest chat platform with about one point three billion monthly active users.
You can also build subscriber list which you can market to with 90 percent open rates if you want to learn more about that.
Have some other videos this one are going to be focused on automating customer service messages so let me show how you let me show you how this works.
Let’s say I’m running a store that sells supplements that help improve people’s memory and focus a customer I might have a question about shipping times.
That’s a pretty common one.
Instead of waiting a few hours to get back to them you can set up instant responses like this one.
So let’s take a look at this response.
There’s three parts here so setting expectations right so it says hey this is the automated helper so we’re not pretending to be human.
This is pretty important because people dont like to be just talking to a dumb human but theyre okay with getting an instant response.
Thats maybe not as custom as the human is obviously going to give them but it is the right answer right now.
So then obviously answering the question and also giving some room there for it to not exactly be the right one because we want to give them an out in case we havent got it 100 percent right.
So then the last part is linked to that and its giving options and making sure that they feel heard.
So by saying you know was that helpful yes or no kinder.
Thats doing some some really helpful things for us.
When it comes to tracking but the main thing is it helps the customer feel heard understood and were not just giving them.
This is the answer.
Now either buy it or leave.
So because were giving these options at the end we can track how well the responses are working so we want to be aiming for that 80 percent of people saying yes that was helpful.
And also how many sales are being made a result.
So we have some some technical stuff here which we do but basically we just add tracking codes to the end when someone’s click yes and then they buy.
So this is using the same natural language processing that Google uses for Google assistant and Google home.
This means that once you train it on one question Ill understand other questions with the same meaning but different wording.
So its not like just a keyword that you enter in and it has to be exactly that thing.
This is its learning all the time.
So if I ask something like will it get here sooner or later how fast is postage.
It will understand that too.
And at first its going to seem pretty stupid.
But over time as you train it more youll find yourself only having to answer around 20 percent of customer questions yourself.
So now let’s take a look at the interface for training and how you can integrate this with your sales and customer support team.
So easing slack for this I mean Slack’s pretty good and it gives a lot of flexibility.
So heres what it looks like when a customer messages your store and the automation doesnt sound you know it just says you didnt understand and you can just jump in there you can reply directly to that customer but then you can also train it so to train it you just click a few buttons.
You either add a new response or select an existing one and the next time that user asks the question the bot will understand.
So over time it will get smarter and smarter.
So Ive just given the real quick overview of this.
If you want to learn more about how to set this up and some of the ins and outs of getting this to work really well then Im going to be making some other videos on this but in the meantime if you want to learn more.
Get this set up for your business.
Just shoot me over an email to Duncan at chat marketers dot com and we can set up a quick call and thanks watching.