Email Support: The Pros and Cons of AI Customer Service Tools (Updated July 2021)

Written by Can Ozdoruk on Jul 25, 2021

Customers today demand more. To meet expectations for convenience, companies need to offer omnichannel programs and email support system solutions. While channels like Twitter, Instagram, Facebook Messenger and WeChat gain prominence, email support will dominate as the digital channel of choice for customer service in the foreseeable future.

According to Hubspot4, 62% of customers want to communicate with companies via email for customer service. This compares to 48% who want to use the phone, 42% who like live chat, and 36% who want “Contact Us” forms. An aside: the best chatbots can work on email (as opposed to just chat and social channels).

Consumers prefer to communicate with customer service 47% more often over email compared to live chat.

An email support system is superior in many ways. It’s not asynchronous and therefore is inherently more convenient. A person can send an email to a company and walk away, checking for a response when it’s convenient for them. A person is not committed to engaging in a real-time conversation which can get interrupted or lost due to connectivity issues or by simply clicking to a new Web page. Email also keeps a record. Emails can be stored and accessed later. On a phone call or Webchat interaction, no record exists.

Learn more about companies using chatbots for customer service.

Why Aren’t Traditional Email Support Systems Meeting Expectations?

Email is the most used digital customer service channel according to Forrester 1, but 62% of companies do not respond to customer support emails2. This presents a real risk to customer loyalty and satisfaction.

For the companies that do respond by email, they are not doing so quickly enough. Customers expect businesses to respond to their emails within an hour3. The average response time to a customer service request is 12 hours and 10 minutes2. Furthermore, only 20% of companies are able to completely answer questions on the first reply2.

From these numbers, it’s clear that email customer support needs to be improved. Customers want resolutions in their inbox. Making this worse, customer expectations for service in every channel are increasing; they expect faster response times and better responses. Companies like Amazon and Zappos have set the benchmarks for customer service, and people now demand quick, convenient support from every company they do business with.

“The handful of companies that respond promptly and accurately to customer emails increase trust in their brand, bolster customer satisfaction, and boost sales both online and offline.”

How Can Conversational AI Impact Your CSAT?

When launching an AI-powered Agent, email is often forgotten. Most companies start with social channels or live chat. Email, however, offers incredible benefits in terms of training an AI and delivering high customer satisfaction (CSAT).

AI Agents managing email conversations need to be more advanced than those deployed on a website or social channels. This is because email messages are typically longer and contain multiple intents. It’s important that you leverage a conversational AI platform that has the ability to decipher the intent from longer messages within the right context to be able to accurately resolve a ticket and reduce frustration.

Conversational chatbot AI can eventually manage over 50% of emails without human intervention, according to what we see from our customers. This offers the convenience and immediacy customers want on their preferred channel which keeps CSAT high.

Discover the key questions to ask when scheduling a chatbot demo.

Should You Launch a Virtual Agent on Your Customer Support Email?

Training an AI is initially done using historic data.

When you’re training an AI, it’s important to not focus on how you think people ask a question, i.e., the FAQs listed on your online help center or knowledge base, but how people are actually asking questions. In one example, Comcast found that their customers ask the simple question “I want to see my bill” in 7,500 unique word and phrase combinations.

By training with historic data, you’re able to set the AI Agent up for success by giving it knowledge and confidence to correctly classify intents across a larger number of utterances. The more data that’s available, the more accurate an AI will be. Because email has been used by companies for much longer than other channels, companies tend to have a much larger dataset.

Email is the most desired and used channel by customer service organizations and their customers.

When providing and omnichannel experience, it’s important to note that your customers ask things differently based on the channel in which they are seeking support. Chat is usually succinct. An email might contain multiple questions and additional detail. In a nutshell, people will ask things differently on email than other channels. So using your troves of historic email data will be most useful if you’re using it to train an email-based Conversational AI.

How Can Email Impact Reinforcement Learning?

People don’t expect an answer immediately on email like they do on chat or voice. This allows companies to start leveraging an AI Agent “behind the scenes” and conduct reinforcement learning. In this type of training, an AI Agent has no direct interaction with the customer. Instead, an AI reviews every incoming email and suggests a response to a human agent. The AI learns how the human agent responds in order to build confidence over time. Because the expectation for instantaneous support does not exist with an email support system, you have the leeway to train with real interactions without disrupting the customer experience.

Excited to deploy AI for your email? We can kick start improving your email channel by free analysis of AI fit in your organization. Just ask us how.