- Define objectives. To choose a chatbot for your business, understand the purpose of implementing the chatbot and the desired outcomes to select the right features and functionality.
- Know customers. Customize the chatbot based on the customer journey, optimize for different platforms and introduce it at appropriate times during user interactions.
- Consider budget. Align chatbot capabilities with budget constraints and prioritize features that enhance customer experience without driving them away.
Even though chatbots have technically been around since the 1960s, in the past few years their use has skyrocketed. And with ChatGPT upping the ante, figuring out how to choose the best chatbot for your business just got a lot harder.
It’s not that choosing a chatbot was ever particularly easy. There are literally hundreds of chatbot providers deploying a myriad of chatbot options designed to meet specific use cases and budgets.
Chatbots can also be very expensive to deploy and train. A poorly designed and executed chatbot can harm the customer experience more than help, so it's important to figure out why you want to use them and where they will be deployed.
Understanding the Different Chatbot Types
The first step in any deployment is understanding the technology you’re looking to deploy. There are two main types of chatbots: rules-based and those that use AI. There are also hybrid versions of these but these are the two mainstays of the chatbot industry.
Rules-based chatbots are the most basic and least expensive type of chatbot. They are used widely to answer basic questions or can help direct people to the right customer service and or sales agent.
Rules-based chatbots come in two types: scripted and button-based. Button-based chatbots basically present users with a multiple-choice menu of options and answers. Scripted chatbots use predefined questions to elicit responses that direct users to predetermined outcomes.
These bots are best used for lead-generation activity and integrate with back-end CRM systems for data capture. They are good at helping users navigate self-service portals and can include interactive elements in the dialogue flow.
On the downside, they are not very personalizable, they do not have natural language processing (NLP) capabilities, and they are only useful to achieve predetermined outcomes.
AI chatbots, also known as contextual or conversational AI chatbots, use NLP and natural language understanding (NLU) — two foundational artificial intelligence technologies — to understand what users need as well as more esoteric things like sentiment. ChatGPT is a conversational AI chatbot.
Advanced versions of these chatbots create conversations with users by keeping track of what has already been said. These chatbots also integrate with CRM and other core business systems to find users the information they need without escalating to a live human agent.
Conversational AI chatbots can respond to complex queries and use machine learning to continually improve both their knowledge base and how they interact with users. On the downside, they can be inaccurate if trained on poor-quality data. They're also expensive to build and operate.
With the rise of ChatGPT, these types of chatbots will be in high demand as people come to the realization that they can now talk with a machine and the machine will give them back useful answers to open-ended questions just like a human would, said Gaurav Kachhawa, chief product officer at conversational messaging platform provider Gupshup.
“Absolutely ChatGPT has raised the bar for all chatbots, and expectations from a chatbot are going to be much higher going forward,” he said. “At a broader level, similar to how the internet drove a revolution in web experiences, and the iPhone app store drove the revolution in mobile app experiences, we believe ChatGPT marks a key inflection point that will drive a revolution in conversational experiences.”
How to Choose the Right Chatbot
Define Your Objectives
Even with ChatGPT and conversational AI now part of the chatbot landscape, the basics of choosing one haven't changed.
The first thing you want to do is understand why you need a chatbot in the first place. Without an understanding of why you want a chatbot and what you are looking to accomplish with the technology, you will not be able to decide what features and functionality will be required to achieve your objectives.
For example, if you are simply looking to capture names and email addresses from an inbound marketing campaign then a rules-based chatbot that directs website visitors to an ebook will likely suffice. But if you are looking to deploy a customer self-service portal that can understand a user's unique set of circumstances and then escalate the interaction to a live agent when someone is getting frustrated, then a high-end conversational AI chatbot that understands user sentiment would be in order.
Some things to think about when putting together the business case for your chatbot include:
- The questions it will most likely encounter and the level of service it will need to deliver
- How much time it takes the chatbot to respond to queries
- How easily it integrates with multiple core business systems such as customer relationship management (CRM), order management, marketing automation and enterprise resource planning (ERP)
- If it will be hosted on-site or in the cloud
- And the level of customization that will be required
As with any technology initiative you'll also want to define key performance metrics so you can measure the success or failure of your efforts.
Know Your Customer
Since all chatbots have to be customized, it pays to understand the customer journey through your systems — be they employees accessing a self-service HR portal or a customer in need of warranty service. In order for the chatbot to be effective, it must be trained on how your organization functions and the data that's required to answer the questions it will most likely encounter.
Depending on the use case, you’ll also want to make sure your chatbot can be optimized for mobile and social media platforms. This enablement can help create a more holistic brand experience.
Another important consideration is for your chatbot to know when it should begin interacting with customers. Few things are more annoying than a chatbot that pops up immediately and demands your attention, said Nat Onions, vice president of customer experience at customer engagement provider Customer.io.
“If I'm on a website for the first time, I don't want to speak to a salesperson,” she said. “I'm just looking at things. Give me a little bit of time to try and figure things out. If I look at the pricing page and the contact page, then maybe I want to talk to someone.”
A good time to introduce a chatbot into the visitor’s experience is when they are looping, said Onions. This is when they click on one homepage and then click on another page and then go back to the homepage and it becomes obvious they are not finding what they want.
Consider Your Budget
While it may be enticing to think about having ChatGPT taking over all the customer service functions on your website, the cost of such functionality may be beyond your reach. Chatbot pricing is typically customized for each client so you should have some idea in mind of what you're willing to spend to get the functionality that you need.
Also most customers still prefer talking to other people over a machine so you don't want to overspend on a chatbot that drives customers away. It's a good idea to ensure that whatever chatbot you choose knows when to escalate the interaction to a live agent.
Consider Your Technology Options
Once you know what your customer needs are and your budget, it's time to start exploring the different chatbot technologies available to you.
When evaluating chatbot providers consider factors such as ease of integration with your existing systems, scalability, security — especially if it's going to be handling personal information of any sort or will interact with international customers whose home countries have stringent privacy regulations — and the level of customization that's possible.
You'll also want to consider the provider's track record, the level of experience with similar projects and ability to provide ongoing support and maintenance.
Finally, don't forget to think about the user experience. Your chatbot should be easy to use and intuitive with clear prompts and responses that guide users through the interaction.