A.I. First Strategy for Customer Service
Artificial Intelligence has finally become mainstream, but in 2017 it’s also become the latest buzzword for business. Everyone from small business owners to major corporations are now proclaiming their products are “powered by A.I.” in order to cash in on the craze, but has the use of all this “A.I.” actually improved the usefulness of the product or the customer experience? Adding voice commands to your B.I. system or implementing a chatbot which can order lunch for your employees sounds great on the surface. But have you considered how this new conversational interface changes the experience, and whether it’s actually for the better? Poorly designed use cases leave many wondering why there is so much hype behind A.I.
Don’t get me wrong — there have been any number of great applications of A.I. that execute well beyond the best a human (or a group of humans) can do. For example, the DeepMind A.I. which improved Google’s data centre cooling efficiency by an extra 40%, saving hundreds of millions of dollars, in addition to being better for the environment. Even with some of the best data centre specialists in the business and very sophisticated technology already being used, the A.I was still able to find a 40% improvement!
What people tend to forget is to get this level of improvement, Google committed a team of A.I. and Machine Learning experts to work with a ton of historical data over two years of development. Sure, looking at the above numbers now, anyone would agree it’s worth the investment… but who could have predicted that amazing level of improvement at the start?
As you probably already know, some of the current challenges around implementing A.I. are that it’s quite costly and it can be difficult to navigate at this point in time. So, how should A.I. fit into your company’s strategy?
We see A.I.’s biggest potential for success in the customer-facing space rather than the back office. Servicing your customers via easily-deployable chatbots can result in a huge uplift in engagement and much higher NPS.
When done well, we’ve typically seen a 30–35% increase in conversions with our chatbots
The reason for starting with A.I. chatbots for customer service is that they can be created quickly and easily, and rapidly iterated up on to create a great experience. They’re also very accessible to customers through messaging apps they’re already using, such as Facebook Messenger, WeChat, etc or the voice devices coming into the market like Google Home and Amazon Alexa, etc.
However, to ensure you deploy a chatbot which will achieve its goal of improving customer experience, there are a number of key considerations you need evaluate first:
Use task-specific chatbots instead of “one bot to rule them all”
Taking into account the current technology limitations, we recommend you take a task-specific approach to deploying chatbots — ensure your chatbot has a very specific use case, rather than trying to cram everything your business does into it. Having a general ‘virtual assistant’ for all potential customer interactions will require the chatbot to ask customers a number of questions to ensure it has understood the context and intent of the their query, and this quickly becomes annoying (especially if it makes a mistake).
For example, consider the word “hot”. Depending on the context used, it can indicate temperature, if something is spicy, popular, sexy, stolen, etc… Humans have an innate understanding of the intent behind the word from the context of the conversation. Unfortunately, the A.I. technology behind chatbots can still struggle to understand the intent when there is a wide context. However, if you narrow the context (e.g. relating to a pizza order, hot = spicy; all pizzas will be hot in temperature when it comes out of the oven) then the capabilities of today’s technology can be put to better use.
So, a task-specific approach will allow for each chatbot to limit the context to just one area or task, and your customers will have a better experience as a result of this minor limitation. This isn’t to say that you can only do one thing with a chatbot — the chatbots can be linked together as well!
Let the chatbot drive the agenda and conversation
The best chatbots are those which drive the agenda and the conversation for the customer. Starting off a conversation with a greeting such as “how can I help you today?” can be confusing for the customer and the chatbot — the customer could reply with anything at all, and if the A.I. can’t make a connection with its content, it will fall back on asking the customer to try again. This is setting the chatbot up to fail right out of the gate; it will frustrate the customer, and it will make them less likely to want to engage with a chatbot again in the future.
If the chatbot drives the conversation agenda by asking relevant questions, the customer can get the assistance they need quickly and easily, and they’ll be able to see the value of the interaction from the first conversation
For example, a chatbot helping you choose the best credit card for your needs would ask a few quick questions (how you plan to use your credit card, when you pay it off, etc) and make a smart recommendation on which product would suit you best. As the chatbot only assists with credit cards, the limited context means that if the customer asks questions about transactions, it ‘knows’ the customer is asking about transactions relating to the credit card rather than loans, bank accounts, trading accounts, etc. Consequently, the chatbot doesn’t need to ask follow-up questions to understand the context, and the experience is more streamlined for the customer.
It’s okay for chatbots to identify as robots!
As yet, a machine hasn’t been developed which can pass the Turing Test over a prolonged conversation.
We recommend that chatbots be upfront about the fact that they are robots
A human won’t take long to work it out, so advising them from the start will help you to manage expectations.
In addition, your customers will be more forgiving when they know they’re talking to a robot — we’ve found when a chatbot can’t handle a request correctly and asks a customer to try again, they are more willing to do so in order to find the answer they’re looking for, without waiting to talk to a human.
If the chatbot continues to have issues understanding the customer, we hand off to a human customer service agent so the experience isn’t compromised further. We don’t see chatbots as a replacement for your existing customer support teams, rather they’re an augmentation that can handle common, repetitive tasks or scripted answers while your customer service team can assist with complex questions or situations requiring more empathy and understanding.
Connect with customers where they already spend their time
Messaging apps are now the #1 place where people spend their time on mobile
They’ve even overtaken traditional social media channels such as Facebook, Instagram and Twitter. Messaging apps such as Facebook Messenger, WeChat, etc, have a number of advantages over traditional websites, apps or even ‘live chat’:
- Persistent Channel — once a user closes a web chat window, you can’t reach out and connect with them again. However, the messaging channel is always open, so you can have an ongoing conversation without having to repeat identifying or preference questions in every interaction.
- No App Downloading — messaging apps now have over 5 billion monthly active users globally, so your customers are just a tap away from interacting with you on these channels.
- Notifications Are Enabled — the majority of users have notifications enabled so they don’t miss messages from friends and family. Your chatbot will cause notifications to pop up on your customer’s phone when you send them a message too!
With a chatbot available on these messaging channels, you’re enabling a great customer experience at scale, without the need for a large team.
Utilise the best features of each messaging platform
Having one chatbot service multiple messaging platforms is convenient from the company’s point of view, but it can provide a poor experience for your customers.
The use of features such as quick replies, image carousels, location sharing, and seamless payments provides an efficient and rich user experience. Cutting out all these great features so you’re only using a couple which are available across all the platforms means that you’re settling on a poor, text-heavy experience for your customers. If they’re just reading text, it’s no different to looking up the information on your website; your customers, especially Millennials and below, are looking for something colourful, fun and engaging to use. You don’t want them to view your chatbot as inferior to the other chatbots they’re talking with!
Chatbots must be engaging, conversational, and suit your use case and company brand
The last, and most critical consideration is the conversational design and tone of the chatbot. These things will truly make or break the customer experience.
A great customer service team member is knowledgeable about your products, their conversation is engaging, and they acknowledge your opinions and frame of mind when they’re speaking to you
Your chatbot needs to be able to do these things too. Replicating a one-to-one human interaction with a chatbot means you need to consider personality, conversational tone, and a personalised experience which matches your organisation’s brand, while staying true to the purpose of the chatbot.
By implementing an A.I. First Strategy for Customer Service, you’ll be able to leverage the best of today’s A.I. capabilities without needing to address the challenges of implementing A.I. into your back office systems and processes. We’ve seen clients able to create proof-of-concept experiments within days, and launch full production chatbots within a few weeks; being able to do this without additional development time for a website or app is an added bonus, allowing organisations to rapidly experiment and provide an amazing customer experience.
