5 Chatbot Challenges and How to Overcome Them

By Allan Stormon — Snr. Director of Content

Chatbots are the new front line for customer service — reducing the impact on human agents and helping businesses save significant money in the process.

However, experiences with chatbots have so far failed to meet expectations. Often conversations with bots can lack flow, they can feel clunky and they often fail to resolve the central issues at hand. While chatbots are still in their infancy, it’s important to understand some of their pitfalls and shortcomings so you can implement a stronger messaging strategy for the future.

Here are some of the main chatbot challenges developers and brands are currently facing along with some tips on how to overcome them:

Problem 1: ‘Bot-speak’ and cold user experiences

Let’s face it, there are thousands of bots available across a variety of platforms, yet very few are memorable. A lot of this has to do with the lack of personality and conversational flow the majority of bots offer. Most interactions are robotic, clunky, and frustrating. Case in point: CNN’s news bot:

The conversation with the CNN news bot deteriorates when the user mentions anything outside the parameters of the programmed script.

While effective at delivering news, the conversation with the CNN bot has little feeling or personality associated with it. When engaging with bots, people don’t need to feel as if they’re interacting with a human, but they do need to feel as if they’re being heard.

The fix: Empathy and simplicity

Before you write any code, place yourself in the user’s shoes. Ask yourself:

• What experiences do you enjoy? What do you hate?
• Are there little things that you appreciate in text-based interactions?
• What specific words or responses make conversations more enjoyable?
• Is there anything you find particularly frustrating?

Use the answers to these questions to map out a bot that provides a satisfying user experience based on empathy. Other quick-fire tips to avoid painful ‘bot-speak’ and frustrated users:

• Inject humor where possible — the more you can make your user smile, the better.
• If it aligns with your brand, use emojis.
• Keep the language simple. There’s no need to over-complicate things.
• Anticipate areas of conversation that aren’t directly related to the bot’s purpose. Program scripts based on these fringe topics, such as weather, current affairs, and basic jokes.
• Slow down — don’t overload the user with huge chunks of text. Maintain a steady pace of information and content that’s akin to messaging with friends.
• Build out the personality of your bot with background information to give it a back story. That way, when the user asks where the bot is from or what it’s favorite color is, it doesn’t reply with “Sorry — I don’t understand what you’re saying.”

The goal should be to delight your customers at every opportunity. Even if the bot fails to solve the customer’s problem, if it can make them smile, your brand can still walk away with the win.

Problem 2: Bots can’t relate to humans

The ability to understand basic language and specific scenarios is a significant issue for bots. In fact, it’s going to be a key differentiator between the good, the bad and the downright useless. For eg. bots that quickly identify a customer service issues and resolve the issue, are going to be far more useful than those that repeatedly ask qualifying questions.

Bots are designed to follow a specific path and for the most part, they rarely accommodate deviations away from a programmed script. Unfortunately for the user, this means many bots can’t understand even the most basic commands or responses if they fall outside of the programmed sequence. This results in a repetitive and unfulfilling experience.

The fix: Build, refine, launch, repeat.

The use of Natural Language Processing (NLP) and machine learning are keys to success here. For bots to get better, they need to be programmed with the ability to learn from the conversations they’re having with users. The initial bot may be limited in its ability to answer questions and drive a conversation with a customer, but over time it will learn and get better.

The technology behind NLP and machine learning is still young. It will be some time before the experiences are as robust and intuitive as we would like. Not all bots can be programmed with machine learning, nor do they need to be. However, it’s important for businesses to start experimenting and investing in the technology now so they’re not left behind when the technology matures.

No matter how simple your first bot is, keep developing and growing it over time. Use the customer data that you gather through bot-driven conversations to improve the experience incrementally.

Problem 3: Chatbots can’t solve everything (yet)

As mentioned, many bots available now are clunky and offer a poor experience. But they will get rapidly better as the technology develops. The lack of functionality in bots is important to consider but it shouldn’t prevent you from exploring how chatbots can benefit your business.

The fix: Add humans

The best and most fulfilling customer service scenarios combine chatbots and humans for a well-rounded experience. There’s no shortage of uncommon inquiries, unique requests, and specific situations that your chatbots can’t handle. These inquiries can be easily handled by enlisting the help of humans to work in unison with bots.

However, it’s important that the transition between bots and humans is quick and painless. When a chatbot is presented with an inquiry they cannot answer, they need to know when to engage a human operator to take over. If this process is clumsy or takes too long, the customer experience suffers.

Problem 4: Bots as another channel for spam

Marketers have a habit of ruining everything. With this in mind, many businesses will be fighting a strong urge to use bots as just another channel to send push notifications, repurposed content, and SPAM through. But this flies in the face of the conversational web.

Bots provide a unique opportunity to develop conversational and interactive connections with customers. Ignoring this opportunity and opting to use bots as one-way promotional tools isn’t going to deliver the kind of experiences customers are seeking.

Bots excel as engagement tools. They are a separate platform to your other marketing channels. The communication that flows through them needs to be fresh, original and unique.

The fix: Build strategically

Identify a specific goal for each bot and structure the interactions around the action you want the user to take. What process or interaction specifically are you hoping to automate and why? What does the user want to accomplish by engaging the bot?

Then, build a bot that achieves these goals no matter how far the conversation may deviate. Script ways for the bot to drive the conversation back on track if it deviates from the original function.

But most of all, respect bots as a conversational platform. Encourage two-way interactions that are equal parts user and bot. When executed well, bots are an exceptional brand-building tool that can drive customer satisfaction and even loyalty. Don’t miss this opportunity by failing to apply strategic thinking and filling your bots with spam.

Problem 5: Lack-luster bots

It’s always a chore engaging with bots that fall short on functionality. Bots need to add value so when they’re not used to their full extent it’s a frustrating user experience.

For example, the Fandango bot takes users to their website when they hit the payment stage. Considering payments within messaging is available, this feels like a missed opportunity to deliver on customer experience for Fandango.

Fandango can increase the functionality of its bot by enabling payments within the messaging stream.

The fix: Know your customers and know your bots.

When designing a bot for business, it pays to have a clear understanding of your customers/users along with what technology is currently available. Use this knowledge to develop chatbots that satisfy customer needs. Don’t lead users through a lengthy conversation without an appropriate end-point. The more functionality you inject into the user experience, the more likely users will engage with your bot.

Here are some examples of what’s achievable when it comes to bots:

• Payments and transactions
• Scheduling and appointment reminders
• Polling and surveys
• Identification and authentication services
• CRM integrations
• Calendar integrations

Bots are the future of customer communication

Chatbots represent an effective and easy way for companies to scale mobile messaging with users. But without a clear understanding of the current pitfalls, you risk building an experience that’s frustrating and useless.

With the knowledge above, you can usher your brand into the messaging era and build a conversational bot that drives results.

Contact us if you’d like to learn more.

Originally published at www.pypestream.com on September 28, 2016.