The four-part plan for 'hiring' the perfect AI customer service agent
The interview with a customer service applicant is underway. After looking at skills and background, it's time to determine whether the applicant will be a good fit. You’ve conducted hundreds of interviews before, but this one’s a little different — you're interviewing an agent that could handle thousands of conversations simultaneously and respond to customers in a personal way, without delay.
This agent works around the clock and doesn't miss an opportunity to upsell. That's the advantage of automation. And despite the differences, just like hiring a human customer service agent, there are several layers to “interviewing” an artificial intelligence agent before you make it part of the team. So do you start by asking for its 30-second pitch? Do you ask where it sees itself in five years?
This sounds absurd, but in an age when 45% of customers prefer talking to a chatbot and 83% expect brands to reply on social media quickly, it's more important than ever before to find and implement the right AI for your business. Conventional solutions to common challenges are no longer feasible, as customer expectations rise and competition for customer loyalty becomes more fierce.
The not-so-artificial challenge
Ironically, while AI is designed to foster high-quality customer experiences, it has the opposite effect when it's not implemented well. Immediate responses, for instance, are only effective when they're customised and thoughtful. Customers enjoy interacting with a machine when it's quick and remembers context, not when it churns out generic responses. Customers want the convenience of interacting with a chatbot without sacrificing a personalised, "human" experience.
That's just one factor distinguishing the good from the bad. In many cases, low-quality automated agents can't adapt to a customer's situation, and they can struggle to provide basic personalised service such as answers to order-related questions or queries related to product selection. Although these questions are simple from a customer's perspective, automated agents often hit a dead end.
Most importantly, good AI agents should know how to use their resources. The AI should know when to refer customers to a live agent, and its technology should make deploying additional capabilities a straightforward process that doesn't require a colossal IT project. Good AI agents might not be able to do everything, but they always stay on script, they never take a sick day, and they're always available across the channels customers use most. Plus, they take care of the most common inquiries, freeing up human agents to focus on more pressing, complex customer service inquiries.
The plan: Navigating the 'new hiring process'
Just like any other job interview, it's important to understand the criteria you're screening against when selecting an AI agent. Look under the hood, per se, to find out whether the AI agent is able to access customer data and understand it well, whether the agent can perform specific tasks or is equipped with skills, whether new capabilities can be added, and whether the AI can seamlessly cooperate with your existing support systems and agents, especially in web chat where many interactions happen.
Now that you have a general understanding of what makes a good AI agent, let's dive into strategies you can use to "hire" the right digital customer service agent for your business.
Start with a gap assessment
This entire process starts with determining your organisation's needs. Is your real-time response currently limited to standard business hours? Can your web chat channel provide instant resolution? Are you also available on the other channels, like SMS text, Facebook Messenger, and Google Assistant? Similar to any other interview, taking the time upfront to assess the areas in which your organisation needs to improve is the first step toward finding the right AI agent.
Then, consider your audience's priorities. Right now, there's a disconnect between what customers want and the ways brands look to serve and engage customers. It is key to recognise that what was previously acceptable is now derided by many customers as inadequate.
Identify common inquiries
Next, narrow your focus to why those consumers are reaching out to you. 57% of customers who contact companies are asking questions, while 45 percent have problems with their products. Customers might also reach out to ask about shipping updates or how to return a product.
Considering that these are prime opportunities to form relationships, your AI agent should provide nuanced, personalised responses that meet the customer's real need in that moment and guide them through resolution, thus enhancing the customer's perception of the brand.
Distinguish between "answers" and "resolution"
Of course, an AI's responses can contribute to a common stereotype — that AI agents just repeat generic phrases to everyone — if the solution used is a nascent one. But automated responses don't have to be generic, nor are they limited to simple questions and answers. Specialised tech and real-time methods of customer data access make it possible for AI assistants to interact with customers like a well-trained representative, using existing business logic and decision models in order to reach a resolution even with more complex requests.
Sephora is an example of a company using a chatbot to go beyond answering basic queries. The brand's Messenger chatbot helps customers set up appointments with a beauty specialist. It also enables customers to hold their camera up to a face to identify what products were used to achieve that makeup look or even determine a product's shade. Customers can even hold up their camera to an outfit to find complementary color palettes and products.
Find an experienced partner
Unless you're fortunate enough to have IT resources in machine learning and AI, consider finding a partner in your search for an AI agent. Typically, this field is not an area that internal IT resources will have extensive experience in, and it's one of the more challenging parts of technology to develop and deploy. Someone who has deployed AI agents for similar brands can help fill in the gaps in knowledge, help develop benchmarks, and provide real-world comparisons that are relevant to your business. Look into case studies so you can set yourself up for a successful partnership and have expectations based on results other brands have seen.
This unconventional "hiring" model is a paradigm shift, but given the advancing technical landscape, all marketing professionals should invest the time to understand where AI fits into their business model. When you survey the results of your new hire a few months after launching, you'll be glad you "interviewed" several digital candidates to find the perfect fit.
Find out more about Digital Marketing World Forum (#DMWF) Europe, London, North America, and Singapore.
- » Instagram best practice for 2020: Focus on comments in a post-like landscape – and try ‘nano-influencers’
- » How to ensure your marketing team's BI dashboards are actually useful: A guide
- » What you really ought to know about ad-blocking users – and why they’re swerving your ads
- » CMO evolution: Understanding the need for change but fearing not keeping up
- » AI for programmatic is great – but tragedy shows us why human judgement remains vital