Start building your own chatbot now >

Up until 2015, social media dominated the digital world, creating a new way of sharing content, managing customer support or chatting around the world. We all got used to it and were expecting instant gratification and simplicity out of everything digital.  In 2015, messaging apps such as Messenger, Whatsapp, Kik or Telegram overtook social media in terms of users.

This created a new shift in the way we get information online: conversation became the new way of sharing, the new interface. That of course had a strong impact on the service industry, which consequently faced a drastic increase in online interactions. Thus, companies today are having to find ways to efficiently manage the increased requests while still providing outstanding customer support.

The industry numbers speak for themselves: 60% of respondents believe that one minute is too long to be on hold, 42% complain about the need to speak to different agents, and 78% terminate contracts because of bad support. This amounts to $1.6tr annual losses due to poor customer support in the US alone. Clearly, something needs to change.

What can be done?

Hire more agents? While that would be the default solution, most major companies currently have thousands of support agents and have reached a plateau in productivity. Hiring new people doesn’t really improve the stats.

Write more FAQs and support content? While that could be a solution, all support departments know that most people don’t bother to go and look for an answer but want to be told directly and in real time. Also, every customer believes they have a unique issue and wants personalized support.

It’s not more agents, it’s not new content, surely the best solution is automation.

But how can we automate parts of customer support to provide instant and interactive answers on different channels? Conversational technology is the answer through the integration of agents powered by artificial intelligence.

"Chatbots are making their way into the landscape of customer relations with the opportunity and promise to offer a better customer experience. The hype is as strong as the benefits this new communication channel can bring: a simple interface, an enriched customer relation with personalized recommandations, a 24/7 availability, etc"
— Olivier Laborde, Innovation & Digital Transformation Leader / Author, @labordeolivier

Now, it’s only natural to feel apprehensive when thinking of AI. After all, “72% of Americans are very or somewhat worried about a future where robots and computers are capable of performing many human jobs – more than double the 33% of people who were enthusiastic about the prospect” (source). However, when talking about customer support automation, we’re not talking about humanoid robots telling you to turn your machine off and on again. We’re talking about chatbots. Chatbots are computer programs capable of understanding human language and to reply accordingly. They’re not physical beings: they’re an interface you can chat with on your phone or laptop.

"For me, chatbots are full of promises, even if they still have a way to go. "
— Jérôme Colombain, journalist High Tech France Info, @JeromeColombain

Today, the goal of chatbots is not to understand everything, rather to handle a selected number of topics, such as invoice management, order tracking, account management. These are simple and repetitive tasks that do not require the added-value that humans bring. By prequalifying requests, managing simple questions and routing customers to the right services, bots also make the lives of support agents much easier.

"A chatbot can efficiently respond to simple customer questions or manage appointments. However, chatbots will not replace apps (unless maybe the most basic ones) or human customer relations. Their strength lies in their ability to reply instantaneously to simple requests and to qualify demands before redirection."
— Camille Jourdain, blogger and author, @camillejourdain

A good example is our collaboration with a major French service firm. When we met with them, they explained their struggle with the increasing number of demands overwhelming an already overstretched customer support service. We sat down together and identified the most frequent conversations with their customers and determined which could be automated. Usually, we focus on automating conversations that can be qualified as simple: exchanges of 3-4 messages where an answer can be easily provided.

The solution put in place was to automate these with a chatbot to provide 24/7 availability while reducing the support workload, following what we call the receptionist pattern. The receptionist pattern is when the bot is designed to understand every user input and is capable of either handling them autonomously or redirect them to the correct agent. In some complex cases, we combine the two: the bot manages the first part of the conversation, usually gathering customer information, and then hands over to a human agent.

Today, the bot is able to accelerate the handling of all queries regarding invoice management up to 2 or 3 times. It works hand in hand with human agents on other queries by gathering client information such as the contract number, the correct phone line involved, which specific invoice or what part of the invoice, before transferring the conversation to the agent. It is also capable of transferring any customer to the correct service on the first attempt, avoiding endless redirections.  

"Chatbots are key tools of good integrated customer experiences. They must not be created in an information island but, whenever possible, integrated with the company’s customer relationship management systems. The easier this integration is, the better it will be to create state of the art chatbots. Also, there must be a human supervision system in place, whether the chatbot is empowering human operators or the other way around. Most of the time, artificial intelligence is powered with human originated data and content. This is particularly true with chatbots. "
— Olivier Ezratty, tech blogger and author, @olivez

The chatbot allowed our client to drastically improve customer support satisfaction by reducing conversation duration by half and reducing the rate of multiple transfers, showing direct impact on their churn rate. The bot also allowed for an increase in the support centers’ productivity: after all, the bot successfully resolves 20% of all conversations. With the time gained, agents also refocused on tasks with a higher added value like sales or personalized support, therefore generating new revenue, something all brands should seek.

This is only an example of what AI can bring to corporations today. According to Adobe’s latest Digital Trends report, “15% of organizations are currently using AI. And 31% said it is on the agenda for the next 12 months”. Will you be one of them?

Want to learn more about customer service automation through chatbots? Reach out to Justine Baron and the SAP Conversational AI team.

To learn more about building chatbots for the enterprise, read our bot building tips!

Olivier Laborde’s quote is a translation from his article Les Chatbots Sont-Ils Intelligents Ou Stupides ? on Forbes.fr


Also published on Medium.

Want to build your own conversational bot? Get started with Recast.AI !

  • Anonymous

    can you bot can understand and differentiate between positive and negative sentence.

    • Justine Baron

      Yes, the bot can reply based on the emotion of the user, whether negative, neutral or positive!

This site uses Akismet to reduce spam. Learn how your comment data is processed.