Should You Be Using A Chatbot For Customer Support?

When chatbots started trending and taking over our social media feeds, I’m sure many of us thought it’s just another phase, as most experiences with chatbots during the early days created more problems than solved. We researched the ‘bot hype’ thoroughly — did customers want to talk to a bot or chatbot? Will the future of communication be geared towards bots, or will we see companies drive more personalized customer support through human interaction? From the outset, I was not sold on bots — every bot I tried, just annoyed me.

Therefore, it was very interesting when research showed us an entirely different picture. A recent study on 6,000 global consumers revealed that 90% of people prefer using messaging to communicate with brands. In addition to this, bots had started to become slowly yet increasingly intelligent over-time, where once a bot had checked me into my flight. 

If 2017 taught us anything, is that consumer expectations are rising, but brands are often lagging behind. Even with the evolution of technology, customer support has typically been slow, not in real-time, and lastly, turning questions from customers into ‘tickets’ and getting back to the customer hours or days later and thus customers have had to go back and forth between support agents. It seems that the process comes first, and the customer second.

How does a Bot fit into this dilemma of rising consumer expectation/experience?

Let’s start off with a scenario:

“It’s 3 AM on a Monday. Maria, a product design engineer, is preparing her presentation of a new ergonomic adjustable standing desk. The weekend flew by, and now, its crunch time; she has to finalize the presentation. Maria opens the product design application, and starts the login process; the application requests for an activation code. Maria frantically looks for the code in her email and cloud storage but finds nothing. She calls the CAD design software company’s customer service number, hoping that they offer 24/7 support. When a recorded greeting responds to the call, Maria starts to panic.

She then notices an option to communicate with a virtual agent on the company’s website. In less than 5 minutes, the bot understands her question and walks her through accessing her alphanumeric code. A wave of relief washes over Maria as she enters the code in her app, and retrieves the needed files for her presentation.”

Offering support to customers in real-time

Keeping ahead of the competition often involves businesses responding to customers instantly with little to no delay. I’m still not surprised on how many companies I come across who are set-up to help their competitors gain their customers — what I mean by this is, they are consciously or sub-consciously letting customers find their own support, and still collecting support issues via contact forms.

Bots are constantly online and respond instantly to customers’ questions in real-time. If the Bot is unable to answer the question successfully, it can either route the conversation to an agent or if the entire team is unavailable, it has the ability to capture the customers’ details and pass the details over to the correct support agent. Acknowledge your customers instantly.

Repetitive questions

If there’s one thing we continuously hear from our customers is, “how do we automate replies to the same questions we receive multiple times a day?”

This is where Bots have really evolved. As a company, we were being asked the two questions below more than 35 times a day.

• What is your pricing?
• Can I refund this product

Without the use of a Bot, agents are constantly answering the same question and leaving customers who require further assistance in long queues, where they usually get frustrated and eventually leave.

At a European telco, a chatbot was used in a pilot program for a set of common customer queries; it resolved 82% of the interactions itself, raising the statistics to 88% of interactions when combined with live intervention by a human agent. This level of performance was reached after approximately, five weeks of training the chatbot.

Sending conversations to the correct support agent

Once calling my network provider ‘T-mobile’, I spoke to 7 different support reps, and yes, I explained my issue each time. I really hope they used that recording for ‘training and quality’ purposes. Yet, I’m sure most of us have experienced this pain. Not all, but some chatbots are able to understand the questions asked and have the knowledge on which support agent would be best suited to answer the questions. This works by using NLP (Natural Language Processing), which understands what is being asked, even if the answer is unknown.

Cost

Chatbots allow human agents to spend their time on addressing more complex inquiries, whilst providing better service and reducing the number of agents required to deal with growing support needs. For many organizations, this has led to them training their support team to deal with more complex inquiries and develops a better-skilled support team to deal with customers that bots struggle with.

Chatbots learn from every interaction

With larger teams, maintaining consistency across the board gets more difficult and in many cases, different answers are provided to the same question. I tried setting up my phone for international calls once, only to get a bill for $955! I got two different answers to my questions, one, of course, telling me, it will not cost extra and the other telling me no it will (which was after receiving my bill, of course).

Nevertheless, bots have proven to be a great tool for maintaining consistency with FAQ’s and allows an organization to analyze if the answers being provided to customers are correct or requires amending.

From the image below, you can see questions and answers the bot is using and also ‘suggested’ lists, which contains questions customers, are asking and may need to be reviewed to boost consistency whilst providing the correct information.

Last thoughts

Overall, chatbots seem to be here to stay, there’s no denying that. Two important factors to consider when implementing a chatbot is, firstly, understand the limitations of a chatbot and what they can and cannot do. Secondly, dedicate a team member to spend 1–2 hours a week to make sure the bot is answering questions correctly and you are aware of any questions the bot is struggling to answer.

Amritpal Dhangal

COO, Acquire.io

Amrit is Acquire’s Co-founder and CEO. His expertise in customer experience and customer service software has made it possible for hundreds of companies to pursue meaningful digital transformation.