Blogs

As customer expectations evolve around AI solutions and, specifically, Chatbots, customer experience leaders and contact center managers face mounting pressure. They’re expected to enhance service efficiency while maintaining a seamless customer experience. Chatbots, unsurprisingly, have emerged as a powerful tool to balance cost reduction with improved customer engagement and overall experience. 

However, the success of a Chatbot implementation depends on more than just deploying AI solutions; this requires a well-defined strategy that aligns with business goals, customer needs, and operational workflows. 

Here, we will look at what it takes—and what you need—to launch a successful chatbot (with a bonus use case at 8x8).

Why do companies need Chatbots?

Let’s start with a “Why”. Why do we need a Chatbot?

• Enhances and scale customer support organization

• Improves efficiency by automating repetitive tasks, FAQs, and free up human agents for more complex tasks

• Promotes cost efficiency, reduces cost per support ticket, and minimizes human errors.

• Boosts lead generation and sales.

• Personalizes customer experience by referencing various behaviors and preferences

Help customer-focused organizations engage at the speed of employee and customer expectations. For example, 8x8's support chatbot, “Otto,” was created to align with some of the above “why’s” and support our business and chatbot strategy. It appears to be paying off: 

• It has improved customer experience by 25%, 

• Reduced escalation to human agents by 16%,

• Optimized costs for further investment in AI tools. 

But we are not resting on our laurels. We continue to refine our strategy to challenge the status quo in the industry.  

How to make a winning Chatbot

Now, let's focus on the “how”—the steps we took to make our Chatbot, Otto, successful.

 An engaging Chatbot conversation makes users forget they aren't talking to a real person. To create this experience, assess your data and determine the most suitable type of bot for your needs. Consider the following:

1. What are the top 10-15 case types/issues customers contact your support team for? Typically, this would be 60% + of the cases/issues created by the customer.

2. To start, take your top 5 customer inquiries and build a decision tree, a step-by-step roadmap covering all possible scenarios for the given topic:

1. 1. What steps do agents take to triage the issue? Have your subject matter experts (SMEs) build this for you.
   2. Are these steps documented in your Knowledge Base (KB), making it more intuitive for customers to follow, learn, and create value?
   3. Before you start creating the workflow, understand how you can build it using your chatbot platform.

From there, you need to determine whether to use a Scripted Bot (which delivers predefined answers and flows) or a Generative AI bot (which pulls data from your KB). The goal is to build a bot that simultaneously creates customer stickiness and engagement. Select the right approach to build the bot based on your easy-to-complex workflow.

Should a contact center Chatbot be supervised or unsupervised?

In regards to a scripted bot, unsupervised or supervised bot, both have their merits:

What's the ROI of a contact center Chatbot?

Now that you know “Why” your organization wants a Support Chatbot and “How” it could be built, let's define the “What”: mainly, what you measure for ROI. What are the key metrics you will be measuring against? 

Containment rate! Wait, I sound like a common Chatbot now. No, that’s not the intent. I just want to prepare your mind because that's what most companies talk about. 

Containment rate is the percentage of chats not escalated to a live agent, a common chatbot metric. However, this alone doesn't indicate success; you could have high containment but low customer satisfaction. To truly understand how your bot is performing and how you’re getting true ROI, you need to look at the full glass:

• Resolution Rate (ARR): The number of conversations fully resolved (with correct answers) by the bot. Your Chatbot solution should have this feature to help you measure ARR.

• Bot Customer Satisfaction (CSAT): Which answers/flows generate positive/negative feedback?

• Contained Conversations: How does the change in contained conversations impact my escalation rate Automated  (to live agent) vs ARR?

• Total Conversations: How is this volume trending? What does it look like compared to the conversations that were contained?

• Deflection: Is this reducing your support case volume compared to ARR? This helps with measuring cost efficiency. 

That sounds exciting, right? But wait—how do I show this off to my boss and tell them a story of success and ROI with some cool graphs and trends? If that's what you’re thinking, we have some mock graphs to help you understand, which you can also tailor to your needs and further polish.

ROI Graph 1: The Alligator effect:

Over the period (monthly or quarterly), record your ARR (green line) and total escalation % to live agent (red line) in a table and build a line graph (on the primary axis) as shown below:

When the green line (ARR) goes over the red line (contained conversations), that's the alligator effect. In the beginning, it will be slow, but you will see this effect over time. To complete the story, you can add CSAT as a bar graph on the secondary axis.

ROI Graph 2: Deflection (a.k.a. The money graph):

You need three data points for this graph: one is the overall support case volume, two is your ARR%, and the last one is contained conversations. For example, if your support case volume is 10k cases and your ARR is 10%, then the contained conversations are 1000 cases—that’s10% of your support case volume! 

Support case volume and contained conversations are shown in a bar chart, and ARR is shown as a line chart. You will notice that ARR goes up and support volume drops over the period, showing an ROI of deflection and cost efficiency.

Here’s an 8x8 secret tip that improves chatbot efficiency

Before I sign off from this article, let me share a trade secret of mine: this is how we improved our Chatbot here at 8x8. The answer is “close loop”. We review our customer comments and workflows weekly to improve the overall experience—quick execution is key here.

Addressing the negative survey feedback is key to maintaining the high ARR and low escalations to support.

• What are the top 3 negative answers/flows?

• What do customers dislike? Is it the wording, the lack of resolution?

By implementing the close-the-loop process in the last 12 months, our chatbot data revealed a significant spike in CSATs, which has doubled over the past quarters. Getting positive feedback like, "Efficient and solved my problem without needing to speak to a technician, thanks!" helps show that we're on the right track.

The 8x8 support Chatbot is successful because we understand the data that impacts the bot's performance and customer experience. This knowledge allows us to communicate with teams and explain how their work contributes to the digital experience. It also raises awareness of how accurate customer data can power the bot with automation to create personalization and drive business decisions. 

ServiceXRG claims the industry standard for self-service resolution was 25%.  According to Master of Code and  Kommunicate the rate of customer service Chatbot implementation is expected to rise to 40-45% in 2025 with the release of Generative AI and Agentic Bots. With a focus on data and enhanced cross-team collaboration, 8x8 can push the limits and surpass industry standards.

Learning more about chatbots to take customer experience to the next level

Contact the team to learn more about how chatbots can help improve your customer experience. If we are already partnered up for your business` Chatbot we will make sure you get all the help you need to ensure that you and your customers get the most out of it. Monthly health check-in sessions with our AI professionals are already factored in the price for you to stay on top of any technology benefit enhancements, bot optimization opportunities,consultative bot use- case development scenarios and general maintenance.

About the author:

Julie Savage is a Digital Experience Program Manager at 8x8. She is dedicated to improving digital customer interactions. Her support agent background helps her empathize with the frustrations of repetitive tasks, which fuels her passion for leveraging technology like chatbots to automate common scenarios and enhance the customer experience.