Benefits and risks of automating customer service

We have all called a company or contacted its online customer service with a quick question only to wait ad infinitum for the appropriate live or virtual representative to chime in. The need for improvement has long been evident. As with other manual processes, automation can ease the tedium, improving consumer experiences to the company's benefit.

Many CFOs are finding that new customer service automation technology can pare the costs of employing live people at contact centres responding to consumer queries and complaints. As these different technologies evolve, it is incumbent upon finance executives to examine the options carefully.

Range of offerings

Chatbots are "speaking and thinking" virtual customer service assistants that reply to customer queries and concerns in texts much like the way Siri responds in voice when asked questions. Providers of chatbots use artificial intelligence (AI) — in this case natural language processing (NLP) — to discern the meaning of a question or statement to provide the most thoughtful answer.

Some chatbots converse in multiple languages, interpreting contextual nuances and the meaning of cultural idioms, making voice chatbots a useful tool for call centres. Oracle projects that 80% of brands will be using chatbots for customer interactions by 2020, up substantially from the 36% using them in 2016.

There are other ways that technology is changing how consumers interact with service centres, and vice versa. A case in point are Q&As that list questions that NLP software indicates are related to a consumer's particular problem. Multiple answers are presented, from which the consumer selects the most appropriate one. A similar automation tool entails more detailed correspondence via back-and-forth text messaging, in which the NLP software interprets the context of each of the consumer's comments and selects an appropriate response.

"There's definitely a need for sophisticated chatbots within the call centre industry," said Shane Nolan, senior vice-president of technology, consumer, and business services at IDA Ireland, a government agency promoting foreign direct investment in Ireland, including customer contact centres. "There are customer segments brought up on messaging apps, text-based interactions, and prescripted customer journeys for which these technologies work perfectly."

Nolan said the amount of research and development going into the AI behind the robotics is "pretty huge", citing leaps in smarter and more powerful NLP technologies. "The market at the moment is ripening," he said.

More than servicing customers

Automating customer service using chatbots also offers opportunities for cross-selling, brand building, and nurturing a closer relationship with customers. Kevin Kelly, founder and president of New York-based digital advertising agency Bigbuzz Marketing Group, provided the example of a chatbot helping a consumer electronics company customer determine which laptop is right for them.

"The chatbot can be designed to ask questions like 'What do you do for a living?', 'How much are you looking to spend?', 'Do you have any kids?', and 'Will you be the only one using the device?'," Kelly explained. "The chatbot becomes another media touch point to express the differentiating value of the brand. This is a much more personal experience than emailing over a list of FAQs [frequently asked questions]."

It's also arguably a better way for a consumer to find answers to their problems. "In many cases today, if they have a specific problem that needs to be fixed now, the company puts them through an obstacle course," Kelly said. "The customer becomes infuriated, spreading their dismay to friends and colleagues, trashing the organisation's reputation and brand value."

Chatbots obviate this possibility, assuming they work as intended. Customers are served more quickly at less cost to the business. "Call centres are becoming quite expensive, and there is a need for automation," said Keith Strier, EY's global and Americas advisory leader for AI.

Aside from the salary of each call centre service agent, it costs a minimum of $4,000 to hire the person and another $4,800 to train them, according to a blogpost from IBM's AI division and platform, known as Watson. Attrition is high in the sector, resulting in spending these amounts again and again.

Alternatively, chatbots that are programmed with industry and domain knowledge are "hired" and "trained" once. Chatbots also are available 24/7/365, "speak" several languages, provide a level of relative reliability, and are never tired or sick. And they'll never get frustrated or angry with a customer.

The cost of building a chatbot ranges from $30,000 to $150,000, according to business-to-business marketing firm 21 Handshake. Despite the high upfront expense, Juniper Research estimated that chatbots across all businesses will save companies $8 billion a year by 2022.

Other side of the coin

Like anything else that appears almost too good to be true, chatbots have their downsides. Aside from the initial expense, chatbots can cause implementation headaches and may fail to perform as planned. Choosing the right solution is complicated. "Some providers have a tendency to put the words 'AI' and 'chatbot' together in a sentence when often there really isn't much in the way of AI going on," Kelly warned.

The tools are still new, with some kinks to be ironed out. Chatbots can backfire disastrously, evidenced when Microsoft unveiled a Twitter chatbot in 2016 called Tay. The company's goal in encouraging customers to chat by texting with Tay was to connect young people in "casual and playful conversation", Microsoft stated. Within hours, Tay's conversations had become extremely fascist, racist, and misogynistic. Tay's replacement chatbot is called Zo and is designed to avoid discussions on politics, race, and religion.

Strier said CFOs intrigued by the benefits of chatbots should not dive in headfirst. Rather they should consider an end-to-end customer service experience, involving messaging apps, prescripted customer interactions, traditional call centre personnel, and the experimental use of chatbots to assist customers at different junctures in the service journey. "AI does not necessarily make this a better solution than other alternatives," he said. "It all depends on the problem the customer is trying to solve."

In many cases, human beings will still prefer another human being to provide assistance when needed most. According to the Chatbot Consumer Report, a 2016 survey of people using a chatbot to receive service, 71% said the robot had failed to adequately respond to their problems.

The bottom line is this: Just because chatbots are available doesn't mean companies should rush out to buy or build them. Yes, they will likely optimise the customer experience at less cost than current service solutions. But sitting on the fence has its benefits. As Strier pointed out, "When it comes to traditional ways of responding to customer concerns, you can't just go from 100% human interactions to zero overnight."

4 tips for venturing into chatbots

Chrissy Jones, a manager in Communications & Member Engagement for the Association of International Certified Professional Accountants, offers this advice for finance professionals who are helping their companies find and use customer service technology:

1. Do your research. How will a chatbot meet your customers’ needs? What problem are you solving? Before you incur the expense of implementing one, you’ll want to ensure it will be used.
2. Find a great partner. New chatbot software and implementation providers are popping up every day. Look for a few that suit your particular needs.
3. Teach your bot. If you spend the time in the beginning testing your bot, you’ll have an easier time maintaining it after you go live.
4. Have fun. Bots are new and exciting and should be approached in that spirit.

Russ Banham is a freelance writer based in the US. To comment on this article or to suggest an idea for another article, contact Drew Adamek, an FM magazine senior editor, at Andrew.Adamek@aicpa-cima.com.