May 15, 2019 | By

Turning Chatbot Skeptics into Believers

As much as we love to talk about chatbots and the customer experience (and we do) – history has made many consumers (and companies) gun shy about using it. Early applications of the tech weren’t great. In fact, they were pretty terrible, resulting in a culture of chatbot disbelievers. Chatbots of days past would often fail because they used a scripted dialogue where every question and response needed to be accounted for. And once it failed, you were, at best, handed off to a chat agent who knew nothing of your “conversation” with the chatbot or, at worst, you were left at a dead end. Companies found that the technology they implemented to help with customer engagements, just made things worse. Not a promising start to the future the Jetson’s promised.

Since then, the technology has come a long way. Thanks to artificial intelligence (AI), businesses can move away from the scripted linear path of yesterday and be fluid and dynamic, more like actual human-to-human interactions.

Despite the maturity of the technology today, there are still some pitfalls companies fall into which can result in poor customer experiences. Here are a few real-life examples of what not to do.

Not Leveraging Natural Language Processing (NLP)

If you are going to go out of your way to make self-service bots available to your customers, you need to make sure it can understand what it is your customers want to accomplish. You want your bot to be more than a glorified search engine. Take this scenario for example. Two questions were ask to the bot. The first returned a single answer, but it wasn’t even remotely close to the question. Then a list of 10 potential results were offered for the second question. Many of thes answers not being remotely close to the intent (also, there should be an attempt for consistent formatting). This is cringe-worthy for customers and will likely result in them still needing to call into a contact center to resolve their issue.

Pushing Consumers to Different Channels

When a customer attempts to interact with your brand through a specific engagement channel, there is a reason. It may just be how they prefer to communicate in general or they chose that specific channel because of some other constraint (time, place, access, etc.). In this bad bot example, the bot did a good job attracting attention, but when it came down to actually solving the problem, it falls phenomenally short. This bot actually tells the user to leave the chat channel by calling or emailing support or going to a knowledge base (which, by the way, linked to a blank page). In my mind, there are two things seriously wrong with this scenario. First, why would a company even bother using a chatbot if it is just going to suggest a different method of resolution? And second, forcing a customer into a channel which they didn’t choose and is less convenient is incredibly frustrating.

Bad UX and User Flow

To provide the best experience for customers, you need to take the conversation flow into consideration – can the user guide the journey and what is the human transition. In this example, the bot fails in both of these. In general, this bot is a glorified IVR. The text input is not taken into consideration – even with keywords. Additionally, there is no way to transition to a human, neither with a text request nor a button. As with the other two examples, this is a very unfriendly customer experience. With a chat interface, there is an expectation that a customer can have a conversation with either a bot or human. Removing both of those is a recipe for an angry customer.


There is Hope

Luckily, these types of chatbot interactions are becoming less and less of the norm. AI technology has come a long way and continues to rapidly evolve. If you’re considering implementing a chatbot (or improving the one you have), here are some tips to follow.

  • Interactions Must Be Conversational and Intelligent. Use a solution which understands the natural ways in which humans converse rather than attempting to deploy a non-scalable scripted bot. To do this successfully, true Natural Language Processing (NLP) is crucial to delivering the best experience.
  • Personalize the Interactions. Leverage data from other business systems like your CRM that can help the bot deliver engaging, real transactions with your customers. Powerful AI solutions can leverage customer context to tailor each engagement to that customer’s specific scenario.
  • Make the Bot-to-Human Transition Seamless. Bots won’t solve all customer questions, there are some problems that require a human hand to solve. For these situations, you must ensure that the handoff from bot to human is as frictionless as possible. It should carry information and context from the bot discussion the customer just to the human so there’s no time wasted getting the agent up to speed.

It’s up to all of us to start ridding bots of their bad reputation. AI-powered chatbots are changing the face of customer service as we know it. Despite the not-so-successful past, this technology has a promising future – as long as humans implement them the right way. It’s time to turn skeptics into believers.


Discuss / Read Comments

Leave a Reply

Leave a Reply

Explore our other AI Insights or Recent Posts.

Subscribe to the Bold360 Blog

Submit your email to get updates right to your inbox.