SOCAP International

The Chatbot Revolution Is Already Here …

Read the full Fall 2018 issue of CRM Magazine here.

… and it isn’t what you were expecting.

If you’ve ever sent a message through the pop-up window on a brand’s site, chances are you’ve chatted with a bot. Facebook Messenger alone has over 100,000 active bots and growing, and while you may have spoken with one, you might not have realized it at first. It’s no surprise that chatbots are becoming more efficient. They present a way for companies to field questions from customers at scale. They can be easily programmed to deliver a few simple replies to queries or escalate the conversation to a human agent if they’re unable to deliver the answer. The question is: Will chatbots replace human agents entirely?
Chatbots have risen in popularity in part due to customers’ expectations for conversational service. We receive instant replies over text, messenger apps and social media from our friends; we expect the same from the brands we interact with. Chatbots can deliver on that expectation, but as anyone who has ever had an extended back-and-forth with a bot can attest, they aren’t quite human. Context, subtext, tone and other uniquely human qualities don’t really factor into the chatbot experience. Communicating with them as you would a real person is likely to just lead to confusion. Clear, straightforward and very literal is the order of the day. Before this changes, the following technologies need to catch up:

Natural language processing (NLP): This is the ability for chatbots to parse your conversation for cues so they can reply accordingly. It’s the difference between being able to actually chat back and forth, versus selecting a few limited options in a glorified menu. However, bots are programmed with a limited number of “intents,” prompts that elicit a specific action, meaning that they work best when there are only a limited number of potential actions customers can take. Bots and their NLP capabilities are limited—broader questions or complex ranges of interactions aren’t what bots are suited for. 

Machine learning: Machine learning algorithms recognize patterns in complex data sets and can draw insight and predictions based on this information. A machine learning framework combined with a robust database should be very effective at adapting to and anticipating change, much like how the human brain works. In theory, anyway. This and adjacent aspects of artificial intelligence (AI) are what push
chatbots towards becoming human, but they still have a long way to go.
alon quote 2
Computer vision:
This is a complex and relatively young field in AI research, but the principle is straightforward: What if your computer could “see”? Instead of just analyzing text, it allows bots to analyze imagery—identifying objects, colors, patterns and more. It will be essential if we want our chatbots to answer objective questions. Seeing and understanding the world is a crucial ability that humans have, and computers just don’t—yet.

However, there are certain areas where chatbots excel and can be a true asset to any company’s customer support team. Chatbots are as good at or better than their human counterparts at certain limited tasks, such as:

Response time: A single chatbot can serve hundreds of customers at the same time, responding almost instantaneously to every single question. No matter how skilled your agents are, they can’t be in that many places at once, and they can’t type that fast.

Sourcing information: If a customer asks which spring jackets are available in their size in green that are waterproof, an agent can look that up—but it will take a few minutes of manual searching. If all of that information is recorded in your database, a bot can do it in a few milliseconds. The same holds true for shipping status, product availability and more.

For all your agents’ virtues, they still need to sleep, eat and do other things beside respond to customers. Chatbots provide easy 24/7 service, they never take lunch, and they don’t even need to be paid a salary. They do what they do best, which gives your agents more time to do what they do best.

Will Chatbots Replace Humans?

If you’re expecting those same bots to replace everything human agents do, then you’ll be waiting for a while. Should that even be the end-goal of using chatbots, anyway? Short answer: No. There are just too many things they can’t do as well as human agents, and it’s unlikely that will change anytime soon.

For example, bots can’t answer open-ended, complex questions like, “Would this look good on me?” Bots can’t reliably supply a great answer or display clear reasoning, as the mixed results of Amazon’s Echo View can attest. They can’t necessarily apply the right tone and sentiment either. Most oscillate between inhumanly chipper and excessively apologetic, but any modulation between is rare. They can’t intuitively understand what’s behind an interaction, and it’s relatively easy to tell that there isn’t a thinking person behind their replies.

While bots can instantly detect sentiment with NLP, they can only respond as well as they’re designed, and it’s extremely hard to program empathy. Without the ability to infer the reason behind questions or create new replies of their own, they’re just still too far from passing the Turing Test.

However, the real chatbot revolution isn’t a complete AI takeover of the retail sector, with agents losing their jobs to a computer. It’s the pairing of skilled agents and well-designed chatbots to do what they both do best and deliver an exemplary experience. Solutions like conversational forms, which disguise themselves as chatbots, are more effective. These act like a bot, instantly responding to your customers’ queries with questions, while getting info from the customer about their account and their problem. They then connect seamlessly to an informed agent, providing them all the relevant info they need to pick up the same conversation and deliver great service.

With these advancements in chatbots and AI, the role of the agent is evolving to deliver a better, more conversational experience. Their role is becoming more consultative and sophisticated. Bots can handle simple questions about returns or updating account information. Agents can put their specific skills as smart, informed individuals to work instead. They can act as stylists for brands who operate in the fashion industry, listening to customers and responding to their needs by applying their own taste and intuition. They can act as subject matter experts who have a real passion for their topic, but can also connect on a personal level with stories of their own collections. They can even build the kind of relationship that allows customers to open up share their private issues and concerns in a way that they wouldn’t if they were responding to a webform.
alon quote
Bots cut out the busy work for both customers and agents, streamlining the experience so they can focus on what really matters—the customer. The chatbot revolution is a peaceful one—bots and humans striding off into the sunset hand-in-hand to deliver great customer service. It means customers get the fast replies they expect and more of the rich human-to-human consultation they value. And it means agents don’t have to copy and paste the same answers to the same questions hundreds of times a day. It’s a win-win, and the revolution is already underway. CRM

Please join SOCAP andKustomer during “Reimagine:SOCAP’s 2018 Customer Care Conference,” Oct. 21-24 in Salt Lake City, Utah, where they will be a presenter. Learn more at

AlonAlon Waks is the VP of marketing at Kustomer. Alon is a marketing and GTM executive having helped B2B and SaaS companies scale and expand their GTM over 20 years. As VP Marketing at 8x8, he led the demand, SDR and segment functions. Prior, he was the global head of marketing at LivePerson, leading demand and content, and also served as the head of product marketing for many years.