9 Critical Decisions Behind Successful Customer Service Bots

Customer service is one of the most popular business applications of automation and bot technologies. Applied successfully, chatbots not only cut costs, but improve customer satisfaction. In contact center environments, chatbots can reduce the Average Handle Time (AHT) per contact, increase First Contact Resolution (FCR) rates, reduce escalations to higher-cost channels, and decrease agent training times.

However, if implemented poorly, chatbots can lead to customer dissatisfaction and brand tarnish. To ensure delivery of customer satisfaction, keep the following points in mind.

Strategy

1. Determine your bot model

There are two models of customer service bots — a “front end bot” and a “bot assisted agent”. A front-end bot is the first line of customer inquiry and typically serves as a conversational version of the FAQs. These bots escalate to a human agents when stumped with challenging questions. The second model of “bot assisted agent” is when an AI powered bot assists a human agent. In this version, the bot provides suggested customer responses for the agent to modify, if needed, and send. Both models have their pros and cons, and is a fundamental decision before implementing a bot.

2. Connect where the the user is

With the proliferation of conversational platforms, a bot needs to be conversant across multiple channels and connect where the customer is. [24]7’s 2016 US Customer Engagement Index found that the most important factor of great customer service is the ability to “contact the company anyway that I want, and get the information and conduct the transaction I need through any channel.” Furthermore, 95% of survey respondents use at least three different channels and or devices to engage with a company’s customer service! Sam Boonin, VP of Product Strategy at Zendesk notes the need to integrate across multiple chat platforms. “Expectations are different in messaging. Customers want companies to reply on their preferred channel — if I tweet you, then tweet back on Twitter.”

3. Create collaboration cross-department

A company implementing a bot needs to ensure that all departments are aware of the chatbot implementation, especially if it’s a solo “front-end bot.” It’s essential that the marketing, operations and technology teams are supportive of the project. For example, when KLM instrumented their new Facebook Messenger support channel, they experienced a spike in customer inquiries. However, they had prepared for this influx by incorporating DigitalGenius’s AI to power their bot assisted agents.

Design

4. Admit that you’re a bot

Based on an October 2016 study, LivePerson found that 80% of consumers want to be told upfront when they are interacting with a bot. Do not try to pass off a bot as a human. After a few conversational back and forths, the customer will realize that they are conversing with a bot. If the brand had pretended it were human, the consumer will feel deceived. Instead, use visual cues in the user experience, avatar drawings, and other subtle indications to show that the bot is a bot.

5. Integrate with existing business systems

Enable your chatbot to access and relevant customer databases, knowledge repositories and core functionalities. Customers will lose their patience if the chatbot repeatedly asks for account credentials or is unaware of prior purchases. Furthermore, a fully functioning chatbot should hook into back-end systems such as bill pay. Bill pay accounts to 30-40% of interactions of calls or chats at a call center, and the inability to connect to this core functionality renders the bot half useless.

6. Instrument an escalation processes

Chatbots need human oversight to handle complicated situations or an especially intense customer. Do not deploy a bot without clearly establishing an escalation process to route customer to live human agents. In this transition, also ensure that the communication history is maintained so that the human agent has the context of the prior bot interaction. Continue building the bot’s knowledge base over time, and the need for escalation will diminish.

Operations

7. Conduct a phased rollout

The bot assisted agent model requires a phased rollout process into the customer service organization. Human agents must feel supported, rather than threatened, by the chatbot technology. For example, start with an internal chatbot pilot with the most creative, innovative and top customer service representatives. If successful, then add the chatbot module to existing employee trainings and finally into new employee training programs.

8. Support with ongoing maintenance

Implementing chatbots in customer service is not a “set it and forget it” process. Instead, a company needs to allocate resources for maintaining and updating content. Company products, promotions, news, policy and more change over time. A team, ranging from a few people to 10 people or more, should constantly review customer-agent-bot dialogues and fine tune the bots for improved performance over time.

9. Track analytics and across all platforms

As with any platform or campaign, track results. Common customer service metrics are average handle time per contact, first contact resolution rates, escalations rates and satisfaction rates. These should be paired with bot metrics such as conversational steps (number of back and forths between the bot and customer), user sentiment and other metrics. Furthermore, as these conversations happen across platforms, ensure that the analytics system can track a customer message across channels.

If your company is considering implementing a customer service bot, keep these critical 9 points in mind to maximize the benefit you get from automated bot technologies.