How AI in Contact Centers Improves Quality Assurance and Response Times

This article is based on a live interview conducted with Miikka Haavisto, Director of Business Development, LeadDesk AI. Watch the interview here.

There are two critical elements to contact center success: quality assurance (QA) and response times. Traditionally, solutions to improve performance in these areas have needed significant involvement from agents and managers, such as through monitoring calls, providing agent coaching, and manually responding to customers.

Now, though, it is possible to harness the power of artificial intelligence (AI). When used correctly, AI has a huge impact on both QA and response times, without increasing workload for your team.

Why Quality Assurance Matters

Through QA, you learn if customers are satisfied and if your team is handling internal processes efficiently. You can use industry standards as a benchmark to see if your contact center is performing at least as well as your competitors.

Without QA, contact centers have no idea if they are doing a good job, unless a customer decides to give them some basic feedback. You may not find out that your customers are unsatisfied with the service you’re providing them until you’ve lost them.

The Key to Improving QA

Improving QA is all about using the right metrics. You need to choose appropriate KPIs for your operation, such as first-time resolution rate, average handling time, and customer satisfaction metrics like CSAT and NPS. Beyond measuring KPIs, it’s necessary to figure out the reason for the results. This means providing agents with the right tools and the training to use these tools.

Contact center managers need to see how their agents are performing and what kinds of conversations they’re having. The traditional way of approaching this has been to manually assess customer interactions: listen back to calls and read through emails or chats. However, this is a time-consuming process, requiring as much as 40% of contact center managers’ and admins’ time each week. Plus, it takes time away from other activities, potentially resulting in a vicious cycle if this reduces the time you have for customers.

AI Solutions for QA

By using AI, contact center managers are able to significantly reduce how much time they spend on assessing past performance. Several types of tools make this possible.

Call Transcription

The most time-consuming task of all is listening to calls. While it is possible to speed up the audio, it’s only possible to do this to a certain extent. Plus, it may be necessary to listen to important parts of the call several times and you could waste time on long sections that provide no valuable information.

It’s much faster to read the transcript of a call because it’s possible to skim the text to find the important parts. An AI tool with a voice-to-text mode automates this process to provide a transcript almost instantly.

Modern solutions do this to a high degree of accuracy, meaning it’s unlikely there will be parts that don’t make sense.

Summaries of Customer Interactions

To make the process faster still, AI tools can summarise customer interactions — there are solutions available for various languages.

This means the contact center manager doesn’t even need to read the transcript to understand what happened during the interaction and to find out the results of the call. The sales agents can also use these summaries, like checking if there is confirmation from the customer about a deal (which is a legal requirement in many places) without needing to listen to the call again.

If any part of the summary is unclear or missing something, the manager can listen back to just that part of the call or find that section in the transcript. However, since modern AI solutions are capable of providing reliable information in summaries from interactions, this is only likely to be necessary if the manager needs to know exactly what the agent or customer said.

LeadDesk’s AI Call Summariser is available to customers wanting to save time in their QA processes. Get it on the LeadApp Store!

Advanced Insights

Transcripts and summaries provide advanced insights into customer interactions. For example, they make it possible to check how often an agent is providing customers with a specific type of information, such as through promoting an offer. It’s also possible to see whether an agent mentioned certain keywords or phrases.

Rather than just trusting the agent or extrapolating the bigger picture from a select number of calls, managers know the exact percentage of calls where each agent is providing this information. The customer service operations manager can use this insight to provide additional training to agents who are underperforming.

Currently, managers can use this feature to provide agents with insights about what actions they should use going forward to improve the performance of the team. In the future, though, AI will be able to provide these insights to agents directly. Immediately after the phone call, AI will be able to provide a summary and pointers of the call.

For instance, this could tell the agent how well the call went based on metrics the company decided were important.

Customer Sentiment Analysis

Another use of AI in quality assurance is in determining how customers feel about their experiences and the brand as a whole. This is possible with natural language processing (NLP), which uses large datasets to understand the intent and emotion behind language. AI is able to analyze all sorts of content, including transcripts of calls, interactions with chatbots, emails, reviews, and social media comments.

Sentiment analysis is more complex than keyword assessments. Rather than just looking for occurrences of particular words, the AI tool uses its language models to analyse the substance of the message. This means it knows when a customer is being sarcastic and can distinguish the meaning of the same word in different contexts, such as in idioms.

Contact centers can use these capabilities to find out when customers tend to be frustrated (to resolve the underlying issues), to discover what customers think about their service, and to determine what to promote and when.

Sales Personalization

Through sales personalization, AI can support contact centers with their promotions. When an agent is talking to a customer is a great time to encourage a further sale — but promoting it at an inappropriate time could come across as pushy or insensitive and therefore have a negative effect on QA.

AI can detect when there is a sales opportunity and automatically provide the promotion to the customer. Even better, an AI product suggestion engine can choose the right products for the customer, either offering them to the customer directly or providing them to the agent to offer the customer.

The Importance of Fast Response Times

Customers today expect a fast service. How long they are willing to wait depends on the channel. For calls, wait time shouldn’t be more than a few minutes, whereas 12 to 24 hours is standard for email. For chat, though, responses need to be almost immediate.

When you’re able to resolve an issue fast and to a high level of satisfaction, you develop loyalty for your brand.

On the flip side, if you’re unable to provide a prompt service, you’ll have unhappy customers who won’t return to you to purchase products or services in future.

Since gaining a new customer costs between five and 25 times as much as retaining a current one, there is a clear monetary value in good customer service.

How AI Can Speed Up Response Times

By bringing AI into the mix, customers receive a response straight away. AI can start providing customer service as soon as your contact center receives a query, reducing the input needed from an agent.

Email Templates

It takes agents a huge amount of time to write personalised emails. There are often multiple questions to address in the same message, which may require the agent to search for different pieces of information.

With templates, it’s possible for a contact center to create emails in advance for all its most common queries. One contact center may end up with hundreds of templates — but this is not an issue when you have AI because the language model can locate the correct ones and even draft the response. All the agent needs to do is make modifications before sending the email to the customer.

To improve QA at the same time, a contact center may like to add special offers into its templates.

LeadDesk has a feature called AI Writer, which drafts emails based on data from similar situations that were used to train the model. In addition to speeding up response times, this can improve quality assurance, as you can ensure each email mentions certain things.

Get AI Writer on the LeadApp Store.

Chatbots

Chat is the fastest-growing channel for customer service, especially among younger consumers. Unlike with email (where agents can only send one message at once) and phone calls (where agents can talk to a single person at a time), chat enables agents to communicate with multiple customers on various channels at once. It’s common for agents to maintain up to around five conversations simultaneously.

Another advantage is that younger customers feel more comfortable with chat than with phone calls.

A downside is companies need to staff chats 24/7 to provide a consistent experience. If customers see that chat is available on your website one day, they’ll expect it to be there when they return in the future. Plus, if many customers want to chat at the same time, they may end up waiting several minutes between messages, which is bad for the customer experience.

Chatbots do two main things: allow you to extend customer service hours to 24/7 and provide automated answers to the most common questions — typically, around 20% of the questions a company receives account for 80% of the chat volume. These two factors combined mean customers receive answers fast. Plus, when they do have a more complicated question, they’re more likely to reach a human agent who is not busy with repetitive questions.

This increases the satisfaction of agents, who no longer need to spend their days on simple questions.

In some situations, an organisation may not have the staffing available to operate a live chat at all. Chatbots mean the company is still able to take advantage of this channel, as the chatbots can answer the simple questions while directing customers with more complex queries to a different channel.

Automate Processes

When a contact center integrates different channels to its backend system (like a CRM), it’s possible to use email and chatbots to automate full processes.

AI can carry out simple tasks like updating client information, postponing the due date of an invoice, or canceling a subscription.

In addition to providing this feature to agents, organisations can provide the functionality to customers through transactional chatbots. Rather than answering common questions, a transactional chatbot carries out tasks for the customer — without needing to involve an agent. For instance, transactional chatbots can purchase tickets, check inventory, process returns, and track shipping.

All this frees up time for agents, keeping them available for other customers and making their work more interesting.

For example, LeadDesk customer A-Lehdet used AI and Robotic Process Automation (RPA) to help customers manage their magazine subscriptions easily in chat. Now their agents only have to deal with 7 out of every 50 chat conversations.

Both QA and fast response times result in greater customer loyalty. The many applications of AI make it a powerful way to improve efficiencies and customer satisfaction at contact centers. If you would like to learn more about how AI can help your contact center, Get a Demo.