What is a Chatbot? The new generation of AI powered Chatbots

1 Sept 2023

8 min read

What is a chatbot, you may ask? By now, you will have come across them, and most likely you will have formed your own opinion on them. We would  be surprised if that opinion is a particularly positive one.

We've all been in the situation where you get presented with options that aren't what you're looking for. Or if they are what you are looking for, eventually the chatbot just sends you to an FAQ page that most likely you've already read. Finally, after you've exhausted all the options eventually there is a button that leads to a contact form or an email address and you press submit and wait.

This is clearly not a great customer experience. These existing chatbots are designed to try and "deflect" queries away from agents, but often aren't really built to deal with the queries that customers actually get in touch about.

So, if these versions don't really do the job, then what is a chatbot for, really? Fortunately there is a new breed of retail chatbots that are built to actually solve customer queries.

What is a Chatbot?

Let's demystify the concept of chatbots. What are they? Why are they gaining traction in various industries? Let's unravel these questions and more.

A chatbot, at its core, is an AI-powered software designed to interact with humans using their natural language. This can occur through multiple channels like messaging applications, websites, or even mobile apps.

Older chatbots have been set up to have fairly limited numbers of paths that a customer can take, rather than being open-ended. Some "chatbots" have essentially just been search engines for help pages, sending you to the help page that it thinks you need most.

The Evolutionary Journey of Chatbots

In earlier times, chatbots were rule-based systems that could only respond based on predefined scripts – for example on. However, today’s advanced platforms like DigitalGenius offer sophisticated bots capable of understanding complex requests and personalizing responses accordingly. They can utilize machine learning algorithms, enabling them not just to understand but also to learn from every interaction, making modern bots far superior compared to traditional ones.

Despite the advances made, some challenges still remain when it comes to creating a smooth conversational flow and avoiding potential pitfalls for users.

In our next section, we will explore these issues faced by traditional models and how recent developments have helped overcome them.

The downsides to old-school chatbots

When it comes to customer service automation, there is no denying the influence of chatbots. Yet these automated helpers are not without their own set of challenges. For instance, many users often find themselves stuck in closed-loop conversations where a bot simply regurgitates previously given responses instead of progressing towards a solution.

"These 'rabbit holes', as they are commonly known, can leave customers feeling frustrated and abandoned mid-conversation."

Limited Flexibility

One key reason behind this predicament lies in how most bots are designed around predefined scripts and decision trees. This severely limits their ability for flexible conversation handling; when confronted with queries outside their programming parameters, they fall back into repetitive cycles rather than intelligently guiding the user towards resolution.

This is partly because brands have deployed chatbots in order to try and deflect tickets from hitting their helpdesks. If a customer can solve their query without speaking to a customer service agent, that's a good thing for everyone.

But brands should be looking at chatbots as an extension of their team. If a chatbot can answer a question that a customer cannot then that's great, but ultimately if the question is outside the scope of a bot, then it should hand over seamlessly to a human agent.

The Impersonality Factor

Beyond technical limitations, though, lies another major downside: impersonality. Old-school chatbots can feel impersonal because they are trying to channel customers down a particular set of paths. Even modern ones that understand more complex requests can feel impersonal if they cannot access information that's relevant to that customer, such as order history.

The new wave of chatbots are overcoming this by doing just that. By connecting the chatbot to the same systems that human agents have, and using AI to extract the relevant information - order history, order status, loyalty points and so on - the experience can be much more personal.

On top of this, the advancement in generative AI can allow more unique responses from chatbots. So rather than having a set number of responses that a bot will cycle through, generative AI can make these responses unlimited.

Chatbot to Agent Handover

The utilization of AI to manage intricate inquiries via chatbots has revolutionised the customer service industry. However, even the most advanced bots reach a limit where human intervention becomes necessary.

It's at this point that an effective 'chatbot to human handover' strategy comes into play. The goal is a seamless transition from bot to live agent when issues arise beyond the bot's capabilities, ensuring uninterrupted customer experience.

This is more than simply sending the customer an email address to contact, but providing a joined up experience where the agent picks up the conversation. Ideally this should happen in close to real time, but this isn't always possible if agents are offline.

The New Wave of Chatbots

Technology is not static. It's constantly shifting, evolving with the digital tide. If you've previously written off chatbots as providing a bad experience, it's time to think again.

Chatbots can understand what customers are asking

Advances in AI such as natural language processing have allowed chatbots to actually "understand" what a customer is asking. We call this Intent Detection, and it is about parsing the meaning from a customer query. Once you have this meaning, then you can start to categorise tickets and build out responses.

Before chatbots would often use keywords to try and work out what a customer meant, but this could miss subtleties in language. For example, a sentence like "My order status says delivered but I can't find it" might lead an old school chatbot to think that "Order status" is the key sentence, and send a link to an order status page. More modern ones will understand that "can't find it" is a more important part of the sentence and act on that basis.

Just by giving a correct response to a query, this can immediately improve customer interactions.

Chatbots can actually solve customer problems

AI can also extract information from different systems and detect what it should be looking for. By connecting to backend systems such as helpdesks, CRMs, order management systems, carriers, and warehouses, the capability of the chatbot can go up massively.

Take the example of someone saying "My order status says delivered but I can't find it". Here an agent would do something like this:

  • double check the order status.

  • check with the carrier for any evidence of delivery.

  • confirm with the customer that they have looked everywhere.

  • Apologise and offer a replacement or refund to the customer.

  • Process the request for the customer.

  • Raise a ticket with the carrier.

Now, all of these things can be done automatically by a chatbot if it is connected to the right systems.

Chatbots can be an extension of your team

Retailers like Antler use customer service automation as an extension of their team, and see these kind of tools as an "extra agent". This is one that can promptly respond out of hours, and even surprise customers by solving their queries on a Sunday.

This means that round the clock, a chatbot can be hoovering up the simpler queries and solving them, allowing the customer service team to pick up what's left in the morning when the clock on.

Chatbots can Streamline Existing Channels

The goal should be clear - automate existing channels rather than adding new ones which could increase workload. By integrating automation through chatbots, operations can be streamlined and efficiency within an organization can see significant improvement.

An example of this approach in action comes from being able to automatically generate return labels. By extracting all the relevant customer information, a chatbot can generate a return label automatically. Just take a look at what On's chatbot can do:

Tackling employee burnout and turnover

Customer support burnout and staff turnover is a major headache for most brands. Salesforce estimate that on average companies have a 19% turnover in customer service staff in any year. Burnout is commonly cited as a major reason for this.

On one level this new wave of chatbots simply reduce the number of tickets that agents have to see on a daily basis. But more than that, they take up the repetitive queries and low value tickets.

As waterdrop put it to us: "By alleviating the burden of repetitive tasks from the customer success team, our agents can focus on the tickets where they can really make a difference. We feel that this is a much better experience for them because they can actually use their skills and capabilities to provide a thoughtful response to a customer, rather than just saying, ‘Here’s a tracking link’.”

So What is a Chatbot really?

Ultimately for retailers, it's only worth investing in a chatbot if it is actually going to solve customers problems. You need to make chatbots work for you and your customers. If they are not enhancing the shopping experience for your customers then they are not worth the investment.

If you are ready to harness the power of generative and conversational AI and transform your retail business's customer service experience, DigitalGenius awaits you. Step into a world where technology meets convenience at every corner; start reaping benefits today!