Artificial intelligence in customer support
Artificial intelligence is rapidly gaining popularity. So we can find its traces in customer support service as well. Is it any good though?
AI in customer service is represented by intelligent chatbots (like those from FlowXO, ChattyPeople, Botsify). Natural language processing algorithms are improving day by day, learning to understand more and more complex requests. Will alive breathing agents be fully replaced by chatbots in future? That’s still doubtful. For simple requests it is possible, but unique and complex requests or super angry customers will still crave for a real support agent. Still, chatbots can be a good addition to customer service outsourcing in terms of cost saving.
How intelligent are chatbots?
Currently chatbot success ratio is about 30%. So over 2/3 or customer requests are answered incorrectly by a bot (this ratio comes up to 90% for tech-heavy products). Recent studies predict the success ratio to reach up to 60% in the next ten years. What this means for companies? That we still need a human to cover up the chatbot, and in the next years this won’t change.
Another point of concern is that bots cannot soothe irate customers, express empathy, be genuine and sincere. They do not evoke WOW effect.
The best opportunity to step into AI land is to add a search option that returns relevant articles from your knowledge base upon request. When deciding on this implementation, you need to consider the number of your articles – such techy addon will require a lot of effort, unnecessary for low volumes. It’s easy enough to browse 20 titles, why bring a bot in.
Next good opportunity is email support automation. Software like SmartAssist (http://www.answeriq.com/ ) distributes tickets among your agents, suggests possible macros and templates for answer. It can also automatically answer tickets based on chosen criteria. There is one problem though: online articles about machine learning clearly point out that we need a huge amount of input data to calibrate such software. Will your business have enough material to tune this intelligent AI? For most of the companies the answer is no.
Are people happy about the innovation?
Living support agents are happy to hear news about AI in support. They expect it to be very useful even today. Let’s say the chatbot can answer about 20% of all requests correctly. Doesn’t look like much, but it’s 20% less load on human agents. In terms of business it’s noticeable savings in time and money. Moreover, chatbots can cover typical and boring requests, leaving humans with complex and difficult cases.
Are customers happy about the AI? That’s a complicated question. If the bot is able to fulfil their request – then why not. Yet people would like to have an option of asking for assistance from real agents when the bot is helpless. They’d also like to know when they talk to a bot, because this changes the attitude significantly. For this cause new design must be created that will become a standard for chatbots appearance.
One of the options here is conversational user interface (like Siri), a hot trend that has been emerging for the last couple of years. Developers today are inspired by this opportunity, and this genuine interest may reshape the way we interact with digital devices.