fbpx

Optimizing Call Center Conversations Using AI

Table of contents

After many years of covering disruptive technologies here at Nanalyze, regular readers will be familiar with our many running jokes. One of these is the term “John in Mumbai” which is used in reference to outsourced call centers. The joke first appeared in a Dilbert strip where all the agents in a Mumbai call center happened to be named John in an attempt to disguise their Indian roots. If you’ve interacted with a Mumbai call center before, you might know why that is. When that clever ruse didn’t work, call centers moved to Manila where people had names that were easier to pronounce. Now, when Maria Santos answers the phone, she sounds like the real thing and everyone’s happy again. Today, we’re going to talk about how call centers might further improve their operations using artificial intelligence.

We recently wrapped up a global tour scouting technology startups. From people-tracking in Saudi Arabia, to retail AR in Belarus, we met with some great startups that are flying under the radar. One stop we made was in Berlin to meet with an AI incubator turning VC upside down (Merantix), a company building the world’s biggest safety event database (Ava), and a startup that wants to transform sales and customer service at scale (i2x).

About i2x

Founded in 2017, Berlin startup i2x has taken in $16.4 million in funding to develop their own Natural Language Processing (NLP) platform that companies use to improve interactions with customers, primarily in the areas of sales and customer service. We sat down to speak with Magdalena Pohl, a Strategic Account Executive who has worn many hats at i2x having been with the firm since the beginning as employee number two. She’s seen the company progress from a concept scribbled on a whiteboard to the full-blown product offering that it’s become today.

The man behind i2x is serial entrepreneur Michael Brehm whose last venture, Rebate Networks, was a startup similar to Groupon where group sales offerings where used to sell lots of products very quickly. In that capacity, Mr. Brehm found it difficult to oversee the more than 2,500 sales agents and customer service associates interacting with his clients in more than 30 countries. That’s when the idea for i2x was born.

i2x uses NLP to analyze calls in real-time or in archive form (many call centers have mountains of call recordings that sit there gathering dust). Two functions where this can be particularly useful are customer service and sales. Call center agents are constantly under pressure to increase sales at a critical touch point – when you’re actually talking to a customer. This is where the i2x software is deployed to assist agents during calls.

Credit: i2x

Using speech recognition technology, i2x can monitor what words agents should or shouldn’t be using during a call, something that can be defined per each client’s needs. Clever machine learning algorithms can also gather information on your speaking style, such as loudness, speaker ratio and the use of filler words, giving agents personalized feedback during and after every call. The tool can even suggest the optimal time to suggest an up-sell. In order to avoid making call center agents feel oppressed by having AI algorithms monitor their every word, i2x uses gamification to help alleviate the mundane day-to-day grind that entails being a call center agent.

One industry where many calls happen is telecommunications. It’s quite a competitive environment where companies are always battling to steal each other’s customers and look towards technology solutions to help them do it. One of i2x’s reference clients is Vodafone, a $35 billion global telecommunications company. In their first proof-of-concept with Vodafone, successful cross-selling attempts increased by over 15% in just weeks by encouraging agents to use particular words and phrases during calls. Today, i2x has multiple clients across the telecommunications industry with an eye to engage other industries as well. But they’re not the only ones playing in this space.

The Competition

Instead of using third-party NLP tools, i2x invested heavily in research early on and built their own. This competitive advantage allows them to customize their NLP engine around any company’s unique nomenclature. But they’re hardly the only company out there playing in this space. A few years back, we wrote about 13 Startups Transcribing Voice to Text Using AI. One of those was Gong.io, a startup that has raised just over $130 million so far for what they call “revenue intelligence.” It’s all about aggregating all the information you have about your customers, deriving insights from it, and using what you learned to sell them more stuff. That value proposition is clear for American companies to understand, said Magdalena. The competitive environment is much different in Europe, where potential clients are very careful when it comes to adapting new technology.

Magdalena also talked about how many startups in this space are addressing very specific aspects of improving call center operations. For example, about three years ago we spoke with a U.K. company called Afiniti that’s now achieved a valuation of $1.8 billion just by connecting the right call center agent with the right customer. All Afiniti does is take incoming calls for a call center and then route them to the agent who would be most effective in dealing with that particular customer. Afiniti refers to this process by the trademarked name “Enterprise Behavioral Pairing,” and back then they were processing around 400,000 calls a day, a number that’s probably much higher than that now. Afiniti claims to generate an average revenue gain of more than 4% for its clients which include T-Mobile, Caesars Entertainment, Sprint, Virgin Media, and Vodafone.

Afniti is just one of many AI-enabled solutions for call center providers, something that falls under the broader umbrella of machine learning for Customer Relationship Management (CRM). Magdalena thinks we’ll see consolidation in the coming years where leading providers look to join forces and crowd out any competitors that haven’t been able to achieve sufficient traction to stay in the race.

Since i2x developed their own NLP platform, the directions they can move in are limitless. They’re also building out their own big data set that can be used to automate particular interactions. However, i2x’s mission remains to “augment humans.” The company puts the end-user of their system at the center of everything they do. Assisting the call center agent in conducting better conversations is at their core.

Conclusion

Some would argue that spell-checking doesn’t make people better communicators, it just gives them a free ticket to ignore learning how to spell. That doesn’t seem to be the case with i2x. They’re providing AI-enabled assistants that make call center agents better at what they do by emulating the best characteristics of high performers. The gamification elements make the job more fun, and cross-selling/up-selling becomes an ingrained habit over time. As call center technology providers consolidate, i2x seems to be in a good position for a potential exit.

Share

Leave a Reply

Your email address will not be published.

  1. The synchronous voice telephone is outdated technology that has one huge disadvantage. Everyone is herded through a low bandwidth channel of one or more voice connections. Someone queuing and listening to music and lies about their call being important for a couple of hours uses as much bandwidth as someone in a sensible, productive conversation. (If the call was important the company would have provided an office number with a human switchboard.)
    Using voice telephony is as sensible as doing the shopping with a horse and cart.
    If companies provided web access to their service, people could queue their messages and not themselves, and get on with something else. The web text chatline system makes a much more sensible use of employee’s time.
    I know not everyone is net literate, but most people have smartphones which are capable of internet access. Those that use it can take the pressure of old fashioned call centres.

    1. These are some good points. There’s always a trade-off between what’s most efficient and what makes people happy. The way things have been going is to stratify – as you suggested – customer service levels based on the value they provide. If you’re a long-time loyal customer, you get to choose how you want to communicate. Everyone else gets online chat only. Those who prefer picking up a telephone will quickly realize how nice it is to use a competent support chat line. Everyone wins – well, everyone except a whole lot of people in Manila right now. Of course, they are simply being freed up to do more value added activities as everyone likes to say.