Customer expectations on how they interact with your brand are continuing to grow. Customers expect to interact over 6 or more channels. How is this manageable?
Great Customer Experience is driven by how efficient, convenient, helpful, knowledgable and friendly you are as a brand across all touchpoints. That’s a big challenge.
A customer is 4x more likely to switch to a competitor if the problem they are having is service-based. Bain & Co.
We also know that the cost of a sale to a new customer is so much more expensive than to one we already have. We also know that customers want a consistent level of service.
73% of customers say that valuing their time is the most important thing a company can do when providing customer service. They also expect to contact the same rep across all channels.
So let’s summarise that.
“I want you to be accurate and consistent, I want to contact you anyway I want, I want you to jump on it and if you don’t sort it out, I’m off.”
Most customers also think feedback never goes to someone who can do anything about it.
How do we solve this?
The biggest advantage we have is that the majority of customers will contact us via a text based channel, that is email, chat or forms. Almost half will also still use the phone, but we can help with that too.
Because customers use text based communication, we can use our AI to help solve the problems.
There are two ways we process information, one is Intent Classification, the second is Text Extraction.
Very simply, if a customer comes on chat and says,
“I’ve been let down, I’m looking for Widget 2021, have you got anything in stock today?”
Our ideal response is going to be:
“Sorry to hear that Bob, yes we do have Widget 2021, and it can be available today on special order, however we also have Own Brand 2018 which is a suitable substitute and that is in stock now at no extra charge. You can order either of those with me now if you like.”
So, that sounds like we would probably make a sale.
That response is purely AI delivered. There is no human involved and the response is instant.
How do we do this?
Let’s look at the customer sentence again.
“I’ve been let down, I’m looking for Widget 2021, have you got anything in stock today?”
We can start by looking for intent.
Let’s highlight that in GREEN.
“I’ve been let down, I’m looking for Widget 2021, have you got anything in stock today?”
So we can clearly see that the customer is searching for something. Therefore, our response is going to include something that we need to find.
What are we going to find? Let’s highlight that in PINK.
“I’ve been let down, I’m looking for Widget 2021, have you got anything in stock today?”
So, we are looking for something the customer is calling Widget 2021. So, can we identify that in our search, is it in our PMI system? Do we have a PMI system? 🙂
If we can find them, are there alternatives? What does PMI say about this product?
But first, what does the customer want us to do with the product?
“I’ve been let down, I’m looking for Widget 2021, have you got anything in stock today? ”
When we break this down, we can see that we are being asked (there is a Question mark at the end of the sentence), if we have the product in stock . There is also a time period we are being asked for, today.
So we are now able to understand the query, the product and the action we need to take.
We can problem solve this with feeds to help the customer with the value transaction. We are not just sending them a link to a product, but we can help inform them of how to solve their problem. This is a big difference.
But, there is more. We should also look at the sentiment that the customer is bringing to us. They have been let down. This is gold dust, because unless they mention us, i.e you’ve let me down, or refer to their own order history, we can assume this is a switching purchase. They are coming to us because a competitor has let them down.Let’s highlight that in yellowish green.
“I’ve been let down, I’m looking for Widget 2021, have you got anything in stock today? ”
We are able to analyse this sentiment in real time. This means we can respond within the context as set up by the customer. We can show empathy and we can personalise the experience. It’s almost conversational. But it doesn’t have to pretend that is a conversation. We should acknowledge, apologise if necessary (in the first person) and then act.
From the customer sentence above then, we are extracting information and classifying intent. The extractions are purple and red, i.e what do they want and when do they want it. The green is the action they are doing, so this is the intent.
“I’m sorry to hear that Bob, yes we do have Widget 2021, and it can be available today on special order, however, we also have Own Brand 2018 which is a suitable substitute and that is in stock now at no extra charge. You can order either of those with me now if you like.”
The yellowish green are the empathetic and problem solving parts of the response. So, we would be able to put this back into the chat straight away.
*Note: we wouldn’t actually highlight them 🙂
We would also be able to write this response to Bobs customer record. This means that if he then emails us (his chat got disconnected, something else happened) we can respond in a similar way. We can then also make a choice if we escalate to his local “human” for a phone call.
Conclusions & Takeaways
Customers are savvy and know that we know, that they know, that we know a lot more about them. They are happy that there is a transaction between how they buy, how they search and the service they require. They know that brands are tracking and they know that this can help them.
Check:
- What is the interaction service that you currently offer?
- Does it include multi-channel, real-time response?
- Can you scale the service with no notice ensuring that you never leave a customer opportunity out there?
AI in customer interaction experience is still in the early stages. The ‘chatbots’ of a few years ago haven’t helped the reputation of the practice, but with new superior technology and understanding of language within Machine Learning, the next few years will again be a step change in what can be achieved.
The three clear wins here are Speed, Convenience and the Right Information.