Comm100 Agent Assist White Paper - Banner Image

White Paper

Agent-facing chatbots are the next big thing

Supercharge your agents with intelligent assistance

Next Chapter Download PDF

Chapter 3

How does Agent Assist work?

Put yourself in one of your new agent’s shoes for a moment.

It’s your first day, you’ve been introduced to your team, signed into your computer for the first time and you’re ready to start learning how to answer customer queries.

For complex queries, it’ll take you a good amount of learning to figure out when you’re making the right judgment call – those queries that fall into the grey area of your organization’s rulebook where a good answer starts with “Well, it depends on…”

But for a lot of other queries, answers are more straightforward. When it comes to getting comfortable with basic FAQs and straightforward inquiries, you’re not so much learning them as remembering the right sequence of clicks to find information or memorizing answers by rote.

Agent Assist works by scanning text-based customer conversations and providing response suggestions stored in internal or external knowledge bases, chatbot responses, and other knowledge resources you already have stored in text form. Agent Assist isn’t meant to replace the agent completely, it’s meant to drive efficiency; the goal is to help the agent deliver accurate answers more quickly. For example, rather than an agent needing to look up a knowledge base article or a user manual, it gets served up to the agent automatically – eliminating the time needed to find that resource link.

Agent Assist isn’t just helpful for new agents. It removes the need for any agent to have to search for knowledge resources, meaning that routine and repetitive inquiries can be completed without feeling like such a chore. It’s also great as a memory-jogger for more complex pieces of product or service knowledge that agents don’t use often, saving them from going to a different system to look it up and reducing the chances that they’ll go out on a limb with incorrect information.

Locating key resources in the agent console has huge benefits – allowing agents to start using internal resources confidently and with speed, right in the window where they work.

Faster resolutions

Statwolf found that regardless of where knowledge resources are located, on average 12.5 percent of a person’s working week, or 11 days over the period of one year, is lost in searching for information – or to put it a different way, roughly 1 dollar in every 8 paid.

Access to information is critical for achieving fast resolutions. But agents often have to work with knowledge resources that aren’t effectively organized or kept up-to-date.

At one end of the scale, your contact center might have already created knowledge materials stored in a single place. However, these resources aren’t typically presented to agents automatically, and the process of moving to a separate system to look for key information eats up time, even if the agent knows exactly where to look. Eliminating the effort of looking is the core value of Agent Assist. At the other end of the scale, contact centers without formalized knowledge resources put the burden of knowledge management on agents. In the course of a day’s work these agents might need to refer to self-made manuals, binders, sticky notes, tip sheets, and other colleagues to find key information.

Agent Assist helps you get that time and cost back by serving up answers to your agents at the speed of conversation, cutting out time spent looking for information. Agents don’t need to search manually – the responses they need are automatically detected from the customer’s query and presented to agents, ready for use in a click.

Agent Assist suggesting answers within the Comm100 agent console

Agent Assist suggesting answers within the Comm100 agent console

Higher accuracy and quality

Contact center knowledge is ever-changing, fuelled by evolving customer expectations, and changes to your products and services. When a change happens, many contact centers provide some quick training to their agent team on the scope of the change and explain how they need to modify the way they explain things to customers. At this point, many contact centers assume they’ve done the most they can, and hope that agents will retain and use the new information they’ve learned.

But the reality is that agents don’t always memorize key information correctly. Incorrect information passed on to customers, directly affects not only customer satisfaction rates but also impacts First Contact Resolution (FCR) by making customers have to contact you again when they find that they weren’t advised correctly the first time. Wrong answers can easily become entrenched in an agent’s mind when they think it’s the right answer.

Giving out incorrect information is also endemic. A recent investigation into Canada Revenue Agency’s call centers found that agents gave wrong answers to customers nearly one-third of the time. Research on FCR backs this up, suggesting that 30% of customers have to repeat contact because their issue was not resolved completely or correctly on the first call.

The stakes become even higher when your team contains new agents who aren’t only trying to absorb information on your products or services but are also trying to learn the right language to use, the goals they need to hit, and what level of service they are expected to provide. New agents, when exposed to customer queries too soon, can make staggering errors. On the other hand, new agents who spend a lot of time learning and training represent paid members of staff who aren’t helping to reduce your contact center’s workload.

The beauty of Agent Assist is that it helps your agents provide correct and consistent information from a single source of truth – helping new and experienced agents alike give accurate, highquality answers, as well as acting as a reinforcement for training and onboarding programs.

Improve your knowledge resources

It’s not called the front line without reason.

Your agents live and breathe customer confusion and challenges every day. They see firsthand where the problems are with your products or services. There is no one anywhere else in your organization who is better equipped to point out the shortcomings and gaps in your organization’s knowledge resources. The trouble is their workflows coupled with your Supercharge Your Agents with Intelligent Assistance 9 expectations of performance leave them neither space nor time to help you fill those gaps.

Agent Assist easily overcomes this challenge by providing agents with a simple way of giving feedback while they’re on the job and handling customer queries. It’s straightforward and virtually effortless for them to identify a gap in your knowledge base, and with a single click to flag it for review and inclusion.

This turns Agent Assist into a powerful tool for implementing a truly living knowledge management process. Even small contact centers can implement processes aligned with the principles of Knowledge Centered Service (KCS) – a knowledge management best practice which promotes the creation, reuse, and improvement of knowledge.

By allowing agents to contribute to an evolving knowledge base, you’re drawing on the wisdom of your whole team to create a knowledge base that keeps pace with change in your business. With Agent Assist you’re achieving that with minimal effort on their part – which is ideal because despite your agents being the primary knowledge users, developing and maintaining said knowledge is not a core part for their role.

The benefits here are huge. Implementing an easily maintainable knowledge resource directly affects the accuracy and quality of information passed to customers and has been proven to provide dramatic improvements to issue resolution, customer satisfaction, agent training times and even agent job satisfaction.

Machine learning in action

While your knowledge resources will expand and shift over time, and your agents’ command of these resources also expands over time, Agent Assist gets better and better. This happens in two subtle yet important ways. First, Agent Assist ‘watches’ which of its suggestions are chosen by agents, and will offer up the most popular responses in the future on the premise that ‘the agent knows best’. Second, as different variations of a question come in and agents choose suggested answers from Agent Assist, its machine learning capability adds that question variant to its list of addressable questions and expands the scope of the associated suggestion. Just as a human agent will strengthen her own synaptic connection between a question and her best answer the more she uses it, so will Agent Assist strengthen its bond between the two. That’s the promise of machine learning in action.

Improve agent job satisfaction

Let’s imagine you’re driving around a new town for the first time, trying to get from A to B, but you don’t have a map or a GPS. For complex journeys, that trip is likely going to be difficult. You’ll need to pull over multiple times, ask strangers for directions, and commit all your mental resources to remembering what turn to take next. Now imagine you’re making the same journey, but with a time limit. If that journey wasn’t difficult enough already, now you’ve added a significant amount of extra pressure.

This tale isn’t too far removed from what agents experience when navigating hard-tofind or inadequate knowledge resources to solve complex customer problems. Fruitlessly searching for answers while their customer is waiting longer and longer for a response, getting more frustrated as time ticks on, is a recipe for agent stress, not to mention customer dissatisfaction. No agent likes to keep their customer waiting for extended periods of time, but if your knowledge resources aren’t easy to find and use (or if supervisors aren’t on hand to help agents out) long waits are often inevitable.

Agent Assist works like a GPS to help agents through tricky customer queries more quickly and more accurately. Your agent sits in the driver’s seat the whole time, and Agent Assist proactively provides them with suggestions on how to resolve the query. Of course, if that response isn’t quite right for the customer or needs some modification, the agent can make changes before sharing. The agent stays in control throughout, and that makes them less stressed and even happier on the job.

Studies have linked a lack of job resources to negative job satisfaction and increased turnover intentions, meaning that simply by making the right resources available you can improve agents’ satisfaction in their roles – reducing stress levels and freeing them to concentrate on making your customers happy.