Want to see how we do chatbot analytics? Join us for a product demo webinar.

Image of 2 men and 1 women in a business setting with text: How 3 digital leaders navigate chatbot optimization blog

Over the last 10 years, conversational bots have evolved to become a critical technology component of a service organization’s self-service strategy. While the investment in chatbots is on the rise, over 50% of CSS leaders say they find low to moderate value in the chatbots implemented. 

Clearly, we feel that there is tremendous value in chatbots – otherwise, we wouldn’t be in the business. But the fact is that many people implement a chatbot, and expect to turn on the switch and watch it give as great as – or better – results than a contact center team. But, that’s not the case. 

We spoke with digital experts from 3 companies, including lululemon athletica and Windstream, who have built self-serve experiences using conversational bots. They share their strategies and secrets for creating exceptional bots, drawing from their experience in managing and enhancing bots for their customer service organization.

You can listen to the complete webinar Building a better bot: Strategies & secrets from digital leaders, or catch up on the top takeaways below.

Takeaway 1: Customers value responsiveness

In today’s highly competitive market, it’s ever more critical to engage customers with quick and easy responses to their questions. Chatbots give brands the ability to get information to their customers without having to search through the website or speak with a person. 

However, customer expectations are always evolving. What worked initially will soon be surpassed as customers desire new channels and new functionality. Throughout this process, it is important to keep the end goal in mind: the customer experience. When considering a new channel, it is vital to understand how customers will use it, considering that voice interactions differ from mobile or web interactions. Conversation design must be carefully considered. The same applies to new integrations – determining which integrations are necessary, and at what point does it become overwhelming to the customer. 

As complexity increases it is essential to have a plan in place for continuous monitoring and analysis to ensure that you do right by your customer and maintain the same high level of responsiveness.

“We are in a self-serve generation” 

~ Ashish Shetty, Chatbot Product Manager, lululemon athletica

Takeaway 2: Chatbot metrics aren’t just about containment 

When it comes to managing a chatbot, there is a tendency to zero in on containment, deflection, or automation rates. But our panelists agreed this isn’t always the right way to measure chatbot success. Setting a goal of 50%, 60%, or 70% containment, may not be right for your business. It’s not enough to have an automated experience that gets abandoned, only to resume the conversation with other forms of communication, such as email or phone calls. Instead, there needs to be a strong emphasis on whether the bot is serving the right experience to your customer

Once you make the shift to include customer experience, it opens the conversation to other chatbot metrics like satisfaction, handle time, or first contact resolution, giving a much clearer picture of whether the bot is actually helping your customers. A bot that struggles to comprehend basic questions, forcing agent escalation or abandonment, is a major drawback. You need to periodically go back to the basics and improve your NLU and content to give customers what they are looking for. Think about investing in NLU training and bot responsiveness to give customers what they are looking for. How many different ways customers can express the same question? Can my bot understand it, and serve the right answer? That investment will be far more valuable than simply trying to contain conversations without escalation

“First contact resolution and overall NPS/CSAT are more meaningful than containment.”

~ Ashish Shetty, Chatbot Product Manager, lululemon athletica

Takeaway 3: Double down on data

In the world of chatbots, one thing is clear: improving performance is key. Yet, for far too long, chatbot practitioners and conversation designers have been left in the dark about what to measure, what to prioritize, and getting access to the underlying data. Moving towards processes that give chatbot practitioners meaningful analytics and tangible access to conversation data, is a game-changer. Given a new understanding of the patterns and trends found within the conversations, they are in a better position to make informed decisions and take appropriate actions.

 The panelists shared their thoughts on how to double down:

  1. Transcripts are key – Transcript analysis is the key to knowing how customers are experiencing the bot. Whereas you may want to drive automation, transcripts are your portal to knowing what they are willing to do with the bot, vs. when they want to engage with a person. Before making roadmap decisions around integrations, use cases, conversation design, channel expansion, voice vs. chat – always start with the data.  
  2. Start early – As daunting as it may seem, knowing how you want to measure bot performance from the get-go will set the foundation for success later on. It’s easy to get your chatbot into production, but if you’re not measuring performance post-launch, such as looking at the transcripts or the right set of metrics, you have no way to understand whether you’re seeing too much or too little containment, or what the customer experience is. This means you don’t know what to start working on to improve the chatbot.
  3. Constant monitoring – Chatbots and customer expectations are always changing. Unlike enterprise software, chatbots need ongoing analysis to keep your finger on the pulse of how customers are interacting with your bot. What is true today is going to change, and by constantly accessing, reviewing, and acting on the data means you’ll be ready to change and evolve alongside customer expectations. 
  4. Data tells you how customers feel – The most important thing you can do is to continue to focus on what is right for your customers, and your contact center agents. If you are thinking about adding channels, introducing voice, or building new integrations always look at the data to know what your customers are thinking about, and this lets you emphasize what is right for your customer.  

“The data helped us understand what to stop doing, and where to focus.”

~ Chase Pavlu, Staff Manager – Software Engineering, Windstream

Takeaway 4: Embrace change 

Chatbot technology and conversational AI have changed dramatically in the past year, and continue to evolve, making it hard to see around corners and come up with a development plan that stretches more than a few months into the future. ChatGPT has shown us that there’s no telling what’s next. You have to be nimble and embrace when things change. But, at the end of the day, whether you’re thinking about being on new channels, apps, use cases or integrations, the most important thing you can do is constant monitoring, constant iteration, and always take into account what your customers want. 

“As a team, we need to be ready to pivot.” 

~ Melissa

Chatbot improvement isn’t a one-off task, it’s a continuous journey

The experiences of these 3 digital leaders underscore the importance of constant refinement and evolution to meet changing customer needs. They show us that with the right strategy, technology, and commitment, businesses can leverage chatbots as powerful tools for enhancing customer experience and driving growth.

Take charge of your chatbot optimization efforts. Book a demo of the Operations Center bot analytics software and start your journey to chatbot success.

Subscribe to our newsletter

Get inspired

Read more insights on the latest in virtual agent analytics and performance management

Subscribe to our newsletter

We use cookies to ensure that we give you the best experience on our website. By clicking “I Accept” or if you continue to this site, we will assume that you consent to the use of cookies unless you have disabled them.