Four Proven Steps to Embed AI in Your CX Organization

Four hours is a familiar and infuriating wait story for consumers. News stories have focused mostly on the frustrated consumer but ignored the plight of customer service agents. A recent report indicated 81 percent of agents experienced verbal abuse by consumers. Adding to this struggle, contact centers face staffing shortages, overworked employees, and a lack of knowledge transfer from seasoned to new employees, making this a challenging customer experience for everyone.

For that reason, virtual or brick-and mortar contact centers have accelerated a focus on artificial intelligence (AI), which offers the possibility—if applied right—to increase productivity of employees, become the horse whisperer of information, and improve customer satisfaction (CSAT) scores.

Having worked in AI for the past 10 years, here's what I've learned that can help you apply AI in your CX organization.

1. Determine the Business Goals

Consumer expectations have evolved and are forcing companies to interact differently with their customers. Those demands require contact centers to scale their communication channels. Putting customers' needs first will help contact centers focus on business outcomes that are key for success. Once goals are established, it's critical that the data used to build your bespoke AI models is not skewed such that models end up with poor predictions.

Consider CSAT as a business goal. The competition isn't other large enterprises, rather the benchmark should be the usability, versatility, and an intuitive user experience that popular consumer apps, such as Facebook, TikTok, or Uber, offer—that both your agent representatives and customers use frequently.

2. Measure What Matters

Measure metrics that cleanly map to business objectives. It can be difficult to determine the most actionable metrics, and some organizations fall into the trap of following vanity metrics, which are a distorting mirror. Tracking page views for an e-commerce business might be interesting, but if the goal is sales, then increasing page views doesn't drive sales.

Average-handle time (AHT), concurrency, containment, and abandon rates are industry standards. What about considering labor-per-hour, or revenue-per-labor hour as new metrics?

Take conversion rates. While a retail-based team will always have higher conversion rates than telephone or digital agents, brick-and-mortar stores are expensive and don't account for the shift in buying habits, technological advancements, and the way consumers want to engage with companies. Digital interactions scale exponentially and cost-effectively. They let businesses look at things differently to better understand the throughput—the number of issues handled per agent per unit of time&mdash. This has profound ramifications on how sales organizations grow their customer bases and requires sales leaders to transform their organizations' approaches for engaging digital-first millennials and Gen Z consumers who want to engage via digital, rather than voice.

To encourage a shift to digital, companies need to provide awesome and intuitive experiences for consumers, reimagined with the help of artificial intelligence (AI). This experience needs to provide better analytics and automation rates and be staffed cost-effectively for consumers who want to operate more asynchronously.

3. Evolve Workforce Management Practices

AI can maximize automation and scale businesses, but it's critical to understand where changes in processes and operations are required. With a change in how a contact center measures success, there needs to be an accompanying change in the way it operates. The right skills and people on the team are also key. It's key to have people who know and think about your customers and the accompanying journey mapping, operations, integrations, or the professional services that support a company.

The best results with AI are born out of allowing systems to do its own work in processing vast amounts of data, and not from configuring all the rules. That's a paradigm shift for groups responsible for implementing and overseeing AI systems, particularly if they manage rules-based technology that is systemically inefficient because it relies on people programming rules constantly. Managing these rules is not scalable. Across many use cases and industries I've observed, there is a learning curve coming from the configuration mindset to adapting to the flexibility of AI systems.

Take call routing. There's an urge to optimize by defining narrow and specific groups of queues, classifying calls into those queues, and setting up granular rules for when it's full or has availability. Defining rigid rules with such specificity can result in an imbalance of distribution. If instead we preserve larger groups of agents and allow the models to learn which agents do well, with which types of customers, intents, and at which points in the day, AI can bring higher levels of global optimization of more distinct areas.

4. Design AI for People

Customer engagement will be a mix of automation and conversation. When a customer is having a service issue that can only be resolved with an onsite technician, AI can take over for the agent and automate the appointment scheduling. Automation can summarize the conversation and confirm everything with the customer. If the customer ends up needing more help from an automated service, an agent can seamlessly jump into the conversation where a human touch will make the difference. AI can provide the flexibility to leverage automation without leaving the customer stuck if the conversation veers off track.

Designed well, AI should make agents' jobs better. It can identify where agents are getting stuck, or when a customer's mood is shifting in a conversation in a good or bad way, so that in the future AI can help move those conversations along in a positive way where the agents feel supported.

Applying these steps will help your organization think differently when embedding artificial intelligence and provide the opportunity to transform business operations in customer experience.

Rachel Knaster is chief product officer of ASAPP.