CS Automation: Front-End Bots vs. AI-Assisted Operators
We discuss the difference between these two styles of customer service automation.
October 9th, 2017
Customer-favourite channels - instant messaging, social media, and webchat - became the focus of most company's digital customer service (CS) a while ago. But as volume continues to increase, companies must ask themselves how they plan to cope.
We believe automation is the best solution. But, there are different automation strategies available to you, and you need to pick the best one to meet your CS needs. We’re going to discuss different approaches to automation, their pros, and their cons.
What are your CS automation options?
Customers continue to put enhanced pressure on companies to perform; the introduction of bots in the CS market makes it not only possible to match, but beat expectations. To that end, there are two automation options which could push you ahead of the competition: front-end bots and AI-assisted operators.
1. Front-end bots are essentially FAQ centres designed for customers to chat with. They offer limited, pre-set responses to a customer’s query or complaint without the need for human intervention.
2. Although they come in various forms, AI-assisted agents are all a combination of human operator and artificial intelligence working together to help the customer. This could be via bots and operators working in tandem, or a bot taking on the initial stage of the conversation before passing it through to an operator.
Neither of these options have a set application, and companies can develop bots as appropriate to them. But is one definitively better than the other?
What makes front-end bots a solid contender?
How your company views CS will dictate your AI strategy. If you focus on cost-savings as a differentiator, then front-end bots look to be better.
They only demand the fixed costs of development and maintenance, whilst AI-assisted agents will have variable operator costs on top. The £3.13 you could save per webchat would be a significant factor.
Without the need for human input, front-end bots can deal with an unlimited number of chats whilst offering a far reduced response time. That's important when you consider that Capterra found 35% of customer service "fails" reported a poor response time as a factor. But, keep in mind that AI-assisted operators also have a faster response time than operators alone.
Currently, it’s looking like a runaway-win for front-end bots. But AI-assisted agents offer the customer a more rounded experience.
Why would a business prefer AI-assisted operators?
First, AI-assisted operators can still save you money. According to the Pareto principle, 80% of queries relate to just 20% of issues. Some forms of AI-assisted operators offer bots that can solve this 80% of queries as a first port of contact. The reduced volume passing through to operators would mean you can cut labour costs. Employing a similar form of automation could save one of our clients £175,350 per week. No sweat.
Reducing the time spent on those pesky, repetitive queries can also lead to a 50% reduction in operator training time. Operators don't have to lose time on simple queries, meaning they can get straight to the tough ones. This reduces the types of queries they need training for.
Additionally, when we first implemented AI, we saw the rate of potential first contact resolution (FCR) double to 62%. Considering every 1% increase in FCR can be matched by a 1% increase in customer satisfaction, that's not to be sniffed at.
Front-end bots require constant attention on their language-processing to ensure they understand customers. This negates some of the saved costs because this would need expensive man-hours. The AI of an AI-assisted operator is monitored as part of the operator-bot combination package; you wouldn't need a separate testing team.
What’s stopping fully automated customer service from becoming top dog?
There’s no denying the fact that, in a few years, customer service could be fully automated. Front-end bots are currently limited by their natural language processing skills. To make this advancement we need even more intelligent software.
Even internet giant Google confirmed 15% of their total searches are unique. This is a search engine which has been around since 1998 and experiences over 9 billion searches a day.
Thanks to human individuality, Google continues to encounter new phrases on a daily basis. Can we really expect a bot to anticipate every possible phrasing of every message that could be sent their way?
Google has been putting time and money into developing a bot which uses neural networks rather than an indefinite number of algorithms. This would allow the bot itself to learn, meaning it can use previous interactions to determine how best to answer its most recent query.
Google has already rolled this out to deal with the unique searches it experiences. But they know there are problems; neural networks are not as easy to follow or adjust. In time, their bot will be able to derive meaning from every message humans can throw their way. Then we will be a big step closer to bots taking over CS completely.
If not applied correctly, front-end bots can go disastrously wrong. Take Microsoft’s “Tay” who, within a few days of being rolled out, was spitting out hate propaganda. Or consider Facebook's recently shut down bot, which created its own language that humans couldn't understand. Our partial understanding of how bots develop their own neural networks, and how we can tweak them as required, means releasing a chatbot as the only port of call for customers presents too high a risk.
Getting the best out of your customer service
Customer service has already developed into a playing field for companies to distance themselves from the competition. But the introduction of AI means there are even more routes to personalise CX to both customer and company needs. Forrester have predicted that AI investment in 2017 will triple, whilst Gartner stated that, by 2020, the average person will actually have less conversations with their husband or wife than they will with bots.
It’s obvious then that AI is the way forward, but which automation path is best – front-end bots or AI-assisted operators? We believe in AI-assisted operators, because they offer the best of both worlds. The AI will pull your costs and response times down without affecting quality of service, whilst the operator can give each interaction the personal touch. Best of all? They perfectly complement each other, meaning no weak areas in your customer service!
If you want to learn more about automation, and how it can be best applied, subscribe to our newsletter below. We're going to be discussing the many benefits of automation in CS over the coming months, and you don't want to miss out.