AI-driven Knowledge Management in the Contact Centre

AI-driven knowledge management in the contact centre
Steve Nattress, product director, Enghouse Interactive

Getting knowledge management right is one of the most important challenges any contact centre undertakes. The Austrian management consultant, Peter Drucker said: “knowledge has become the resource rather than a resource…”. Certainly, knowledge management is at the heart of the contact centre’s role, bridging the gap between customers’ enquiries and the facts, which will resolve their problems. But it is not just an important challenge, it’s a very difficult one.

Today, with extensive product lines, specialised customers, fast-changing market requirements, and complex partner relationships, there are huge volumes of data to deal with. Added to this, businesses need to think about how fast information becomes outdated, as products update continually and the knowledge generation process moves further away from those selling to and supporting the customer.

Given all this, for any business, finding the right information and getting it to the right customer quickly can be daunting. Maintaining an up-to-date knowledge-base is often the first challenge. In today’s contact centre, the latest AI technologies can help organisations achieve this goal. Whether it’s collection, distribution, or delivery of knowledge, AI can support it.

It can, for instance, automate information delivery through self-service bots; discern the point at which escalation to a human agent is appropriate, and ‘surface’ the right knowledge to help them deal with it.

AI can even apply enhanced semantic reasoning to help the agent work out what the customer needs by making internal knowledge accessible through the terms and language the customer themselves is using — thereby overcoming misunderstandings that lead to dissatisfaction. The level of intelligence being applied in this area is advancing all the time. There are subtle things the technology can increasingly do already: from parsing common misspellings, to understanding different local phrasing, like Hoover being used as a generic word for a vacuum cleaner in the UK, or gas being used in the US as an equivalent to the UK petrol. It works by being a bit fuzzier around the terms people are using, to cut through to what they really mean.

Getting the training right

AI- driven knowledge management is also increasingly key in agent training. The approach can, for example, make a huge difference in the direct delivery of customer service, with one of Enghouse’s customers — selling a vast inventory of white goods’ insurance products — cutting their training and onboarding pathway from eight weeks down to a fortnight for new agents.

They set them a challenge on day one to go and find the answer to questions — and the trainee gets immediate feedback and reward. This way also, the business can quickly embed the approach in the culture, that agents faced with difficult queries need to go and ask the AI-driven knowledge base because the answer they are seeking is going to be in there.

Once those new recruits are actively delivering customer service excellence alongside their colleagues, the consistency and quality of the service is enhanced too. It’s had a knock-on impact on their call handling time, leading to a greater than 20% reduction in percentage terms, or about 30 seconds reduction per call. Whether the agents are working in a co-located or distributed manner, the AI supports each of them in an individual and timely way: overcoming the distance and lack of shared experience, with direct access to information and support.

The customer benefits from AI-driven knowledge management in this context also., even if they may not be directly aware. All the customer knows is that the person they speak to has immediate access to the information needed to understand and solve their problem as quickly as possible, and find the best way to make them happy… So, that’s a win-win, for the business and their end-user.

Looking further ahead

In the future we expect to see, AI tools applied more and more to analyse inbound interactions to ensure knowledge bases are stacked with the most frequently asked questions and customers can self-serve to get what they need.

This kind of analytical semantic capability will become ever more important over time as the capability of the technology grows. Being able to extract content from much larger documents and knowledge bases, and adapting that information for the audience at the right moment, is where the future lies. In the years to come, we will increasingly see instances, for example, of contact centres extracting and formatting complicated product reference information needed by a new customer, from a technical paper into the answer they need right now.

The latest advanced AI takes all this a step further by really understanding the nuances and intent behind language, and collating and surfacing genuinely unmet needs, in ways that customers may not directly express them. For example, if customers are asked to answer a survey that allows for open ended responses, it can be quite hard to read the sentiment of tone of their replies. Their response may be short and unexplanatory. By using interaction analytics you can get at the real concerns they want to fix, then suggest intelligent solutions to solve them.

It’s this level of detail that turns a reactive overview into an action plan. You need to know the WHY, behind what people are saying, and what’s really happening. Once the AI offers the why, all developers will have to do is build what they know customers want — to ensure complete satisfaction.