To Bot or Not to Bot? Getting Automation Right

Of course, the answer is “to bot,” but presenting customers with self-service options needs to happen at the right time — and for the right reasons. Be sure that leveraging a bot and automation for the customer journey provides a great customer experience — and that it aligns with your intended business outcomes.

Artificial intelligence (AI) in the contact center is still in its infancy and with that comes a lot of exploration and trial and error. And while there are countless tools for Natural Language Processing (NLP), the goal must be the same: maintain a frictionless customer experience to ensure customer loyalty, foster brand promoters and increase sales.

The Customer Contact Council found that 94% of customers that had a “low-effort” experience will purchase again, as opposed to only 4% of customers who had a “high-effort” experience. Implementing an isolated bot to reduce calls without considering how it fits into the overall customer experience ecosystem likely will make your call reduction plan backfires.

Here are some practical steps for creating and managing automation and bots in your contact center.

Set Clear Business Objectives

New technologies are almost always trialed in the lab. They’re cool, and companies need to keep up. But rarely do cool lab projects affect a company’s bottom line; the goal of a proof of concept for a cool project is simply to prove that the technology works.

Now that bots are creeping out of the lab, consider the business objectives they’ll support. Determine if there’s a corporate objective for the contact center to contribute to revenue growth, if there’s a particularly costly activity you can automate, or if there’s a push to improve brand perception or Net Promoter Score.

Consider Domains and Intents Holistically

You can’t have a new technology without new terminology. In the bot world, think of domains as lines of business and intents are customer journeys or what customers can accomplish within an line of business. A bank would consider domains to be deposits, loans and mortgages. Under those domains are intents — directives like “get rates” or “apply for a loan.”

Take a step back and develop a strategy for the entire automation ecosystem across business domains. A customer that has a loan and a checking account with the same bank doesn’t care that these services are managed by different, siloed teams. An interaction within one domain is an interaction with your brand; your customers expect a similar experience each time they engage.

Not every bot built across the company needs to be planned out before you start. However, think about the overall automation strategy, how it will affect your brand image and how it will affect the customer journey. Begin with the end in mind.

Prioritize and Map Your Journeys

Once you have a good understanding of what the automation ecosystem will look like, the next step is to determine why customers reach out to you. Then identify which of those reasons have repetitive outcomes and a low effort to automate.

You’ve likely thought big, but you need to start small. An FAQ bot might not be the sexiest place to start, but if a large number of customers contact you with questions that can be answered easily, start there. Use the cost savings from this use case to fund the next effort. A well-designed bot can automate more than 50% of the interactions it was designed to handle.

Once three to five intents are prioritized based on perceived value:

  • Revisit journey maps that include these intents or spend time mapping the journey from the customers’ perspective
  • Determine where in the customer journey the bot fits and how the interaction will gracefully move to another resource — whether it’s machine or human
  • Evaluate customer personas and decide if there should be different bot experiences for different types of customers and scenarios

A bot doesn’t have to be an all or nothing proposition. You might have high-value customers, for example, that you want to route straight to a specialist. “The top 5% of your customers bring in 30% of your revenue,” estimates Custora. The technology is there for these customers to have an experience as special as they are to your business.

Release to a Control Group and Monitor

Be careful. Your brand reputation and loyalty are at stake. It might sound dramatic, but automation and bots shouldn’t be released into the wild until you know they’re doing what’s expected — without sacrificing customer or employee experiences.

Deploy a new bot capability to a controlled group in a logical communication channel. For example, if the first foray is an FAQ bot, employees are the perfect incubator to harden the knowledge and experience before exposing this capability to customers and additional channels.

While narrow AI solutions like NLP engines are making progress on automated tuning and training, you need to supervise their algorithms to ensure they’re serving the intended purpose and not doing more harm than good. Automated utterance detection provides control and simplifies supervised learning tasks.

Introduce Capabilities to Additional Channels

Once it’s clear that a new bot has been sufficiently trained and is performing well, make it available to other interaction channels. For example, if you’ve built a chatbot to process purchase exchanges and have observed the expected affect to customer satisfaction and interaction volume, make this exchange bot available to voice and social channels.

A solution that orchestrates and manages these bot interactions, along with existing digital and voice streams across communication channels like Genesys Designer or Genesys Intelligent Automation, lets you build bots once and make them available to additional interaction channels. And with an open AI platform, you can use the NLP engine of choice or multiple NLP engines.

Don’t Let One Bad Touchpoint Ruin the Entire Journey

Consumers have many choices; how they feel about a brand is often as good as their last experience. Even relatively happy, long-tenured customers become annoyed by a poorly designed application that didn’t enable them to complete their journey within the channel of their choice.

AI-powered customer experience will change the way we do business. And no matter how legitimately tantalizing the rewards are, don’t jeopardize your customer experience and brand loyalty by jumping to the end and building a bot in isolation.

To learn how to design customer journeys that are effortless and improve the customer experience, read our ebook “Journey mapping: An outside-in approach to delivering great customer experiences.”