If you are on Spotify, you’re probably familiar with the Discover Weekly playlist. Every Monday, you get a playlist with new music, based on what you usually listen to. Not a week goes by without a great find. It seems like Spotify really knows you! Discover Weekly made my Monday mornings so much better — until one day it didn’t.

For weeks, my Discover Weekly playlist was loaded up with French hip hop, which —
pardonne-moi — I don’t really care for. At first, I thought it was someone at work playing a prank on me, Spotify-jacking me by following a ton of terrible playlists when I wasn’t around. I unfollowed all those French artists and hoped that by the next Monday, my Spotify algorithm would be fixed. But a week went by, then two, then three, and I was still saying goodbye to the weekend to the tune of Tout n’est pas facile.

This really started to put a damper on my Monday mornings. I reached out to Spotify Support, but they couldn’t help me. There wasn’t a single FAQ that addressed my problem. At one point, I even toyed with the idea of switching to Apple Music.

As it turns out, Spotify’s artificial intelligence (AI) engines were working fine, it was just my account that had been hacked by some random French hip hop fan in Bordeaux. Everything’s back to normal now, I have my perfect Discover Weekly playlist back, and I’m once again a very happy user. But this made me realize two things. First, my hacker had terrible taste in music. Second, when you pay for a service that doesn’t meet your expectations in some way, it can really get on your nerves and you want it fixed immediately, with the least possible effort.

Enter customer support

This is what Customer Support Operations do. They go by many different names — Customer Support, Customer Care, Customer Service, Customer Happiness, Customer Experience, and so on. But they all focus on the same thing: solving customer queries as fast and as effectively as possible.

Many companies have that need on a global level, which means a higher number of demanding customers or users, more timezones to support, and more languages to communicate in. In order to be able to respond to such demand, companies try to structure their Support Operations by splitting them into several contact centers or service hubs, strategically located in different timezones, preferably in low-cost regions where specific language resources are easier to find.

Global companies want to have these contact centers operating as efficiently as possible, and for that they are constantly looking into different ways to reduce costs, increase productivity and improve service levels. Regardless of the industry, every support team can identify patterns in customer contacts: there are types of queries that are much more complex and urgent than others, and in order to address those as quickly as possible, companies try to leverage technology and data to free up agents’ capacity from repetitive, simple queries.

When technology meets contact centers

More and more companies are looking into tech solutions and tools that can help automate customer contacts, so that they can use human resources to address only the most serious customer issues.

When looking at different channels of customer support, each of them can benefit from trends in AI — particularly automating a specific set of tasks and queries based on frequency. Below are a few ways of improving efficiency through AI innovation.

Knowledge Base platforms

When I typed “Discover Weekly” into Spotify’s Help Page, it didn’t turn up anything useful. And I’m not alone. 40% of customers say that help center searches don’t generate the results they’re looking for.

Knowledge base platforms are the best-known form of self-service support and a very good way of deflecting traffic away from support agents, especially for simple requests. But they’re also not the easiest thing to manage. Customers’ needs and expectations change, and new issues come up over time. It’s virtually impossible for agents to track all keywords researched by users and to keep all help center articles up to date while still giving timely replies to customers.

The secret to having a solid self-service platform is in the content. You need to make sure it addresses the right questions and uses the language and keywords that customers search for.
Good news: there are AI tools that can help you know your customers better and fully understand why they contact you and how they think when looking for an answer.

For instance, Google’s Conversational Topic Modeler uses AI to look at a history of written conversations to uncover insights that can help tailor knowledge base content to what customers tend to look for — think “how do I reset my password?”, “why haven’t I gotten my refund yet?”, or “where is my order?”, but probably not “my account was hacked by a French hip hop fan, what do I do?”

Zendesk’s Content Cues works in a similar fashion. The algorithm identifies recurring customer questions across the system and suggests new content to add to your knowledge base and identifies articles that could use updating.

AI is a great tool to enrich an existing knowledge base, which in turn increases customers’ self-service possibilities and helps reduce operational costs.

Chatbots

Contrary to popular belief, chatbots are not here to replace agents; they are a useful tool for agent augmentation.

First of all, chatbots don’t need to sleep. They can cover for nightshifts or lunch times, relieving the pressure of rotating shifts or reduced breaks, while at the same time providing 24/7 support and improving customer experience.

These virtual agents, like IBM’s Watson Assistant or Intercom’s Answer Bot, can also deflect traffic away from the service teams. They feed off of the help center articles you have in your knowledge base to answer the most common and repetitive customer queries, leaving agents to deal with the more demanding ones. So, the more carefully curated your content is, the more likely it is that you’ll be able to solve a part of your queries using this automation tool only.

Chatbots can additionally use AI to learn from historical chat conversations with customers and identify patterns in customer questions, improving the accuracy with which a bot can respond to a query, as well as knowing when to ask for more clarity or redirecting them to the right (human) agent.

Live chat and emails

Usually, live chat and emails require a real person answering queries. We’ve come to a point where automation tools have, for the most part, succeeded in bringing in only the more complex queries to this level of support. But even here, there are ways of using AI to enhance agents’ performance. For instance, there are tools that enable email triggers, templated answers, scheduling appointments or following-up on past orders, allowing agents to fully focus on the queries they have in hands.

At this level of support, queuing is usually not only based on first-come-first-served, because it also depends on the language the customer speaks. Often customers are allocated to a queue according to the language they speak, and the waiting time fluctuates depending on the availability of that language resource.

However, there is an increasing trend towards a Universal Queue with a single waiting list, and certain AI tools like Unbabel enable that by empowering English-speaking agents to respond to customers in any language, at a native level.

Phone

Despite being the most expensive support channel, and also the most time-consuming, it is still necessary, especially for very complex issues.

AI voice tools are still in their early stages, but are increasingly necessary and developing quite fast. Automatic Call Distributor tools (ACDs) help filter and redirect queries to the right agent in an automated way, by using speech recognition and human interactions. There are also many IVR systems that allow callers to access information without having to speak to a person.

And finally, call recording software coupled with AI speech analytics tools can also help to collect data on all customer interactions, allowing managers to evaluate their teams’ performance as well as gather customer insights.

Making contact centers a better place

Customer support teams have typically suffered from disorganization and a huge workload. Some companies have upwards of ten contact centers, either to cover the biggest number of timezones possible, or because they need local operations assured by native agents.

Technological innovations, in particular advancements in AI, are not only helping businesses reduce costs and increase agent productivity. They also offer the possibility of centralizing operations, since times zones and languages are not longer an issue. Centralized customer support teams are, in turn, easier to hire, manage and train.

AI itself can be applied to many different areas and functionalities, leveraging data to predict and automate certain actions. However, we should keep in mind that it does not replace humans, it simply empowers them to do a better job and not waste time on boring, repetitive tasks.

Discover Weekly, for example, is the product of an algorithm that saves me a lot of time looking for new songs, giving me more time to enjoy good music and have a much better Monday. At least until the next hacker comes along.

Source: https://unbabel.com/blog/contact-center-ai-innovation/