Virtual agent chat bots are becoming an increasingly popular tool in contact centres. In this blog we take a look at what a virtual agent is, how it works, the benefits it can bring to a business, and finally, we offer advice on what aspects to look out for when looking to buy virtual agent software. We’ve got a lot to cover, so let’s get started!

 

What is a virtual agent?

It is an artificial intelligence (AI) based system that provides automated customer service. It is employed by many contact centres to provide fast responses to customer queries and/or field initial questions before directing the customer to a human customer support agent.

 

What’s the difference between a virtual agent and a chatbot?

The difference between a virtual agent and a chatbot is not always clear, and we find that many people use the terms interchangeably.

The easiest way to differentiate between the two is to look at their capabilities; virtual agents are typically more advanced than chatbots.

Chatbot – a chatbot is designed to handle basic communication tasks. It is programmed to respond to questions with pre-approved answers using scripted decision trees. When a customer enters a query, the chatbot picks up keywords and responds with the most suitable scripted response from its knowledge base. In a lot of cases, chatbots require customers to choose responses from set phrases (e.g. to start the conversation the customer chooses between ‘account’, ‘order’ or ‘billing’), which limits the scope of user input and therefore the questions the chatbot can be asked.

Virtual agent – is designed to interact with a customer. Rather than relying purely on scripted answers, it uses machine learning and natural language processing (NLP) to understand a customer’s intent and respond in a more personalised, ‘human-like’ manner. An advanced virtual agent can also ‘remember’ previous interactions, which means it can learn to handle more and more customer queries.

The table below provides a brief summary of the capabilities of each system, showing clearly how they differ.

 

Capabilities	Chatbot	Virtual Agent	Notes
Natural Language Processing	✗	✓	Virtual agents use NLP to analyse text and make sense of it, whereas chatbots need exact keyword matches in order to respond.
Understand customer intent	✗	✓	Chatbots typically require an answer from a predefined list of suggestions, which means they rely on keywords and don’t pick up intent like virtual agents.
Understand sentiment	✗	✓	As virtual agents analyse text, they can pick up sentiment (as well as intent) and adjust responses accordingly.
Machine Learning	✗	✓	Virtual agents can learn over time by ‘remembering’ past interactions. This offers huge potential for dealing with organisation-specific conversations.
Decision tree routing	✓	✗	Decisions trees are typically associated with chatbots, but virtual agents can also use decision tree segmentation to simplify a user’s journey.
Scripted/pre-approved answers	✓	✗	While virtual agents draw information from a knowledge base and can only provide answers they’re ‘trained’ to answer, their responses are not scripted word for word like chatbots.
Draw information from knowledge base	✓	✓	Virtual agents are more advanced as they can grow their knowledge base by ‘remembering’ past interactions.
Answer FAQs	✓	✓	Both systems are best used to answer repeated queries, such as FAQs.
Answer complex queries	✗	✗	Virtual agents can answer more unusual and complicated requests than chatbots, but they’re still best at dealing with repeat queries. Both systems will need to hand over to human agents when complicated questions are asked.
Offer personalised interaction	✗	✓	Virtual agents don’t use scripted answers meaning they can personalise responses. They generally sound more ‘human-like’ than chatbots.

 

How does it work?

A virtual agent draws on its knowledge base, machine learning and NLP to answer customer requests in the most appropriate and accurate way it can.

When a customer sends a message, the virtual agent will analyse the words to determine sentiment and intent. Based on this, it will use the information in its knowledge base and experience from past interactions to respond.

In this way, it needs to be ‘trained’ to respond how you want it to. Do you want it to simply direct a customer to the right department? Do you want it to handle FAQs from start to finish, minimising human input wherever possible? Do you want it to gather initial details to save time and then handover to a human agent? There is no right or wrong; you determine how you want it to function to best serve your business.

If you do want it to handle complete customer requests, bear in mind that the quality and amount of information you provide it with will affect the quality and range of answers it can provide, and even then it will have limitations. Virtual agents are great for dealing with routine queries, such as FAQs, but can get thrown by complex issues, which is why a human agent should never be too far away.

 

 

Key benefits for organisations

It has many use cases and can bring plenty of benefits to a business. To help you determine whether your contact centre needs to invest in a virtual agent, let’s take a look at the benefits in more detail.

Improve customer experience – It can help to improve customer experience and satisfaction in a number of ways:

 

• Fast responses: one of the most important factors in determining a good customer experience is a fast response to queries. It can answer customers instantly, making them a reliable way to provide fast responses. ‍
• Reduce contact centre queues: both phone and live chat queues are a big turn off for customers. In fact, 60% of customers aren’t even willing to wait on hold for just one minute! One of the best ways to reduce contact centre queues is to provide a virtual agent. As they deal with FAQs and simple queries, it puts less demand on other channels and significantly reduces, if not eliminates, queues.
• Gather customer insights: They are a good way to gather information and data on customer behaviour and journeys. What are the most frequent questions it answers? What webpages trigger the greatest number of interactions? Insights such as this can show you where in the customer journey your customers encounter problems. You can then work to improve these areas and offer pre-emptive customer support. ‍

‍Reduce demand on agents –  It can handle many customer requests, it is a great tool for reducing the demand on human agents in a busy contact centre. This will give agents more time to deal with complex customer problems and improve employee morale. (We do not suggest a virtual agent replaces human agents, but it can be a very useful complement to human-powered customer support).

Round the clock support – being able to offer 24/7 customer support with a virtual agent is a clear benefit for your contact centre. An automated virtual agent can operate without human input, which means it can take over all customer support during your closing hours, and is also a great option if you have customers in different time zones.

Generate leads – They can collect customer information, either via pre-chat forms or through the conversation itself. This information can then be used to retarget customers in the future with nurture marketing campaigns, updates and product promotions.

 

 Best practices

If you’ve invested in a virtual agent, how do you ensure you get the most out of it? The following tips are essential to making sure your virtual agent does its job and helps you deliver excellent customer service:

Build a decent knowledge base – even though a virtual agent doesn’t completely rely on scripted messages, its responses are still drawn from its knowledge base. This means you need to provide it with the necessary information. If its knowledge base doesn’t contain the right information, the virtual agent cannot provide the best answers to customers.

Integrate with existing applications – like any new software you introduce to your contact centre, it’s a good idea to make sure your virtual agent is integrated with your CRM and other customer service tools. This makes sure information is easily shared across tools (which adds to the virtual agent’s knowledge base) and makes it easier for live agents to take over from the virtual agent if needed.

Enable escalation to other channels (e.g. video and cobrowse)– a virtual agent is not capable of dealing with all customer queries, which is why it should be programmed to hand over to a human agent when it cannot deal with a certain request. There are different ways you can trigger a hand over (e.g. keywords, time passed etc.), but the key is to make sure the escalation is seamless and fast. Key channels for escalation are video chat and cobrowse. These offer customers a personal experience and ensure swift resolution to problems, which is essential if a virtual agent has not been able to resolve the issue.

Go omnichannel – offer your virtual agent where your customers want it! They aren’t just for your website, implement it across the different customer contact channels you offer, such as a mobile app or other messaging platforms.

Proactive interactions – virtual agents are a good way to initiate a conversation with a customer. For example, if a customer has been inactive on a web page for a couple of minutes, a virtual agent can pop up to ask if they need any assistance. This proactive engagement is a good way to generate leads and guide customers to the next stage of your sales funnel. Set up rules to determine when your virtual agent is triggered (remember not to overdo it, virtual agents popping up as soon as you enter a web page can be very annoying and turn customers away!).

Automation – make sure your it is programmed to respond to customer requests and search its knowledge base without human input. Automation is key to reducing human involvement: without it, a virtual interaction will still need to be manually triggered by a human agent.

Multilingual support – if you serve customers in different languages, ensure your virtual agent can offer multilingual support. This can be achieved through real-time translation technology or by adding scripted FAQ answers in different languages to its knowledge base.

Compliance – like a live web chat interaction, virtual agent exchanges need to follow the necessary rules to be compliant with your industry. For example, if it gathers data, it needs to be GDPR compliant, and virtual agents for healthcare providers need to be hipaa compliant. Make sure it ticks all the compliance boxes before it goes live.

This blog post was first published by Talkative.