Article | July 1, 2020
Chatbots have come a long way in the past few years. The improvements in technology have enabled developers to expand on bot capabilities far beyond just functioning as a FAQ. Today, the automation of chatbots can process orders, perform financial transactions, make bookings, and much more. (Check out other intelligent functions here.)
However, as intelligent as bots can be, no chatbot can handle and resolve all your customer queries. It simply cannot answer the infinite number of questions a human may throw at it. The technology is simply not there yet, and it may never truly get there. But perhaps more importantly, brands shouldn’t want a bot to manage every customer query.
A bot working independently of human involvement won’t always deliver the best results for customer or agent. It’s the combination of chatbots and human agents that takes customer service to new heights. What you need is a smart and efficient way of translating your organization’s unique customer service philosophy into appropriate action so that every question is met with an answer in the best way possible – whether that be by bot, human agent, or a blend of both.
To deliver this, you have to pay attention to the who, what, when, and where of customer engagement. You need to know who your highest-value customers are so you can always route them to a human agent, for example. You need to know what they need help with so a simple question can be managed by a bot. And the list goes on.
Here’s why humans need chatbots, and chatbots need humans – and how you can achieve this perfect balance to deliver support that will exceed customer expectations and generate substantial ROI.
Why humans need chatbots
There’s no doubt that supplementing customer-facing roles with automation can yield fantastic results. The launch of McDonald’s self-serve kiosks is a great example of this. By giving customers the option of ordering their meal through a kiosk, or through a cashier, McDonald’s demonstrates the success you can achieve by combining automation with human. Here are just some of the benefits it brought to the customer and employee experience:
1. Automating large portions of simple queries so workers have more time to focus on other, more complex tasks
2. Reducing monotonous, repetitive queries to improve employee experience
3. Catering to customer preferences – choose quick automated service or deeper human engagement
4. Reducing queue times, in turn improving customer experience
5. Lessening the opportunity for human error
6. Generating ROI by reducing staff numbers
These results almost identically mirror the benefits that intelligent chatbots can provide customer service teams. By implementing a bot, a large portion of frontline support can be automatically managed by the bot which:
1. Gives agents more time to handle complex questions
2. Reduces the monotony of answering repetitive questions
3. Allows customer to choose between chatting to a bot or an agent
4. Reduces wait time and queue length (through bot’s ability to handle infinite simultaneous conversations), in turn improving customer satisfaction through quicker resolution
5. Eliminates human error in data entry
6. Generates substantial ROI through lower service costs
See how closely those benefits match?
Recommended reading: Chatbot ROI Calculator
Why chatbots need humans
The relationship between bots and humans isn’t a one-way street. While agents need bots to provide more effective and efficient support, bots need agents to provide the personal, ‘human’ touch that many situations call for. In our latest 2020 Live Chat Benchmark Report, we found that chatbots handle 68.9% of their chats from start to finish – although an impressive stat, it still shows that many queries require an agent’s touch.
Recommended reading – 2020 Live Chat Benchmark Report
There are always going to be situations that call for human assistance: canceling a subscription, reporting a lost or stolen credit card, or registering a serious complaint. Or maybe the topic is sensitive, and your customer would feel more comfortable explaining their situation to an agent. Similarly, some (though increasingly less: stat?) people are still wary or reluctant to communicate with bots and prefer to only speak with a live agent. To cater to these customer preferences, it’s vital that these customers can be routed past or transferred from your chatbot to human agent without effort and without having to repeat themselves.
It’s important to note however, that transferring from bot to agent isn’t always just in the interest of the customer – it can often benefit the customer service team too. This is because not all queries are equal. For example, if a customer reaches out asking about a bank’s opening times, this can be easily managed by a bot. However, when the same customer asks about a loan, this high-value interaction may dictate that – according to your unique customer service view – a human agent takes over immediately to ensure the customer receives the best experience and you close the deal as quickly and effortlessly as possible. If your chatbot can’t do this, turn it off and find a chatbot that can (we can help with that).
How to create the perfect chatbot – human (agent) balance
To begin creating the right balance between chatbot and human, you need a bot that’s widely accessible to today’s digital-first consumers; your bot needs to be where they are, wherever they are. Comm100’s AI Chatbot can serve customers on web, in-app, Facebook, Twitter, WeChat, WhatsApp for Business, and SMS. You also don’t need to build separate chatbots for each channel. Simply select the channels you want your bot to be available on (hint: all of them!) and you’re off.
Although your customers will know they are speaking to a bot (and you should make this clear to them to set expectations), you need a bot that understands natural human language. Comm100’s AI Chatbot harnesses the world’s most advanced NLP engine so that it can understand your customers’ goals and provide the answers they’re looking for. Better still, add a large range of off-the-shelf integrations to this, and the Comm100 bot can begin performing actions on behalf of your customers – from tracking an order and paying a bill, to booking a flight.
By resolving a large portion of your frontline customer service questions, your agents will have more time to focus on higher-value queries and customers that matter most to your bottom line.
Recommending reading: Comm100 Chatbot Resolves 91% of Assigned Live Chats for Tangerine
As we’ve discussed earlier, there will be times when you or a customer would rather connect with an agent than a bot. It’s crucial that your bot offers this flexibility.
Firstly, your bot should be able to give the customer the option to speak to an agent at any time. Eighty-six percent of consumers believe they should always have the option to transfer to a live agent when dealing with a chatbot. You can easily set this option up within the Comm100 AI Chatbot.
Next, you need a bot that can automatically identify the conversations that you want an agent to manage. This requires training your bot on the topics – ‘intents’, in bot lingo – that your customers will bring up. If there are specific intents that are of high value to you, you can tag them so when a customer mentions it, the bot recognizes it and automatically transfers the chat to the appropriate agent or department. The bot can also be trained to notify an agent or escalate the conversation when asked a question it can’t answer or if a visitor is clearly frustrated. As a failsafe, your agents should also be able to monitor bot conversations and take them over in these situations.
Chatbots will never replace whole customer service teams, and nor should they. The ‘human touch’ is still essential to customer support, and we are a long way off until this changes. However, if implemented intelligently, bots can resolve a great portion of customer queries without any human involvement, allowing team sizes to reduce, or remain the same in the face of increased support volume.
Take Tangerine, an Australian telecom company, for example. They experienced rapid growth, which in turn produced a surge in chat requests. By implementing Comm100’s AI Chatbot, up to 91% of assigned live chats were resolved by the bot without any agent involvement. As a result, Tangerine could manage the increase in chat volume without hiring and training more agents. And when high-value customers reached out, their agents were free to provide them with the best experience.
Article | March 1, 2020
Artificial Intelligence (AI) is a powerful force that is already reshaping our lives, environment, and interactions. It may be defined as a program whose aim is to produce human-like cognitive processes and potentially even improving on them. AI has many facets: it may be algorithmic as in game-playing programs or take a control-theoretic approach as in autonomous vehicles. It may also manifest itself as linguistic ability, creativity, spatial reasoning, learning, and many others. We are now recognizing that AI has begun to make huge inroads to Life Sciences, be it making discoveries from huge biological data using machine learning, combining health records and genomic data of various types, discovering new drugs or drug targets, finding new groups of cell types, making diagnosis, or customizing health procedures as in precision medicine.
Article | March 17, 2020
In discussions around the future of AI and cyber-threats, we often wonder when we can expect to see malicious or offensive AI attacks in the wild. While we have not yet seen conclusive evidence of execution, this report will show that all the tools and open-source research needed to facilitate an AI-augmented attack exist today. This report will document an end-to-end attack lifecycle, and how each stage could leverage elements of the AI ‘toolkit’ to improve and streamline the process. Attackers will, of course, evolve their tools to drive efficiency gains, however these tradecraft improvements are iterative and do not happen all at once. Furthermore, while it is likely that adversaries today are already leveraging AI in some capacity to improve individual attack phases, this report shows an end-to-end AI-driven attack purely as a thought experiment.
Article | August 3, 2020
Two of our talented data scientists, Yong Liu and Andrew Brooks, recently showcased how Outreach helps sales reps leverage machine learning for their own continuous learning.At Spark+AI Summit 2020, they talked about how machine learning powers our sales engagement platform, as well as how they solve some of the typical challenges all data scientists face when building enterprise-grade applications.