Global Retailer Assesses Their First 6 Months with Virtual Agents
Electrolux, the 2nd largest home appliance manufacturer in the world who sells under a variety of brand names worldwide like Frigidaire and AEG, made the transformation to AI-powered virtual agents six months ago to automate more in their contact center. Ross Tudor, Head of Electrolux’s Customer Engagement Center, shares their rollout, results, and best practice tips for other contact center leaders.
Q: Why were you considering AI-powered virtual agents to automate more in your contact center?
A: “Customer expectations have changed regarding self-service, and our live agents were simply handling far too many routine calls with defined business rules that required little to no judgment. At the same time, our IVR was limited on capabilities and the traditional approach of hiring professional services to build out more functionality was both time and cost prohibitive for us. We knew if SmartAction’s conversational AI solution could do for us what it has done for others, we would be able to automate a large chunk of our call volume quickly and affordably to save costs and elevate our live agents to critical thinkers instead of doing the same mundane repetition over and over.”
Q: Which call types did you consider for automation and why?
A: “SmartAction helped us identify common call types and processes that had defined business rules within the conversation flow and didn’t required complex critical thinking on the part of the agent. Since the SmartAction model allowed us to start small with very low risk, we identified three high volume conversations that contained a lot of repetition and redundancy:
- Account Verification
- Product Authentication
- Customer and Product Registration
We handle thousands of these call types and processes every single day. Since they follow the same specific process consistently, they were the perfect place to start for conversational AI automation.
Because Electrolux operates many large brands, authentications and registrations are a large part of the basic functions that agents perform every day; in fact, each agent handled these call types around 35 times per day on average. We wanted to make their lives less redundant, and offload as much time as possible from the average handle time (AHT).”
Q: How difficult was the implementation and initial pilot?
A: “Since SmartAction delivers its conversational AI solution via the cloud, the integration was not difficult. Call types or processes intended for SmartAction virtual agents are routed to the SmartAction cloud where the call or process is either completed or passed back to our live agents for the finishing touch, along with a screen pop of gathered data. The only integration work was connecting to our customer database, so SmartAction virtual agents could access and record the same data as our live agents.
We piloted the solution for 90 days, which led to our official launch five months later, which included the full back-end integration, conversation flows, and deeper customization of the solution to our business needs and use cases.”
Q: What do your six-month results look like?
A: “We have maintained a detailed report of which calls were fully contained within call automation, which were not, and why. The virtual agent was required to:
- Fully verify customers using name, zip code, phone number, or other personal information
- Fully authenticate products using model numbers, serial numbers, date of purchase, or other information
SmartAction’s conversational AI solution successfully captures and contains about 60% of calls received from customers who are already in our database. The rest are unable to be contained, due to a variety of reasons—most of which would cause difficulty for our live agents, as well. Here’s the breakout:
- 60% of callers are successfully contained within virtual agent automation
- 14% of callers provide information that does not match information contained within Electrolux’s databases, which can be attributed to transfer of product ownership, address changes, or other such updates. This mishap likely would cause an agent to pause before diagnosing the issue. While this may be something the AI self-service will handle in the future, we simply haven’t included it in our application scope for Electrolux yet.
- 14% of callers request a live agent transfer. For many of these calls, the customer is actually missing some piece of information that is needed. What we like is how the virtual agent attempts to fix a customer’s problem by asking questions in a slightly different way, requesting a different data point, or helping customers locate serial numbers before transferring to a live agent.
- 9% of callers hang up. This can be attributed to realizing they need to be near their appliance to make progress, so these callers ultimately call back. At least now they did not waste agent time while searching for this info.
- 3% of calls were transferred to live agents because the virtual agent was unsuccessful. I should mention all of these calls are being analyzed by SmartAction’s CX team to determine how to tweak and tune the application for future success.
Successful call automation is an ongoing process. Containment rates continue to improve as the SmartAction “AI brain” continues to learn and the SmartAction CX team examines and tweaks. We had a very narrow scope for our first set of calls, but have worked closely with SmartAction to develop our roadmap for future automation as goals are met like they are today.”
Q: How have virtual agents impacted AHT for live agents?
A: “Now that we are 6-months in, virtual agent AHT is just under two minutes. At the same time, our live agent AHT has dropped by almost 2-min for a near 1:1 correspondence. When you’re talking about handling hundreds of thousands of calls, that really adds up.”
Q: How have your live agents reacted to the addition of virtual agents?
A: “Now that we are 6-months in, they’re very positive. Initially there were some trust issues. Agents would receive the baton only to verify the same information, so we had to do some educating. Prior to automation, our agents had to robotically handle the same repetitive call 35 times a day, five days a week. This drove them crazy. By outsourcing routine calls to virtual agents, the live agents can elevate their work to critical thinking that solves problems and delivers the high-value customer engagement we need from them.”
Q: Tell us about the vendor selection process for your conversational AI solution.
A: “When we started our journey, we did extensive research into a variety of solutions that were available. We wanted the best AI technology available but also needed something that was easy and low-risk to implement. After all, we have never done this before. We weren’t going into anything that required a large up-front investment. Our goal was to keep the first foray into AI automation small and targeted to keep risk at a minimum. SmartAction was the only vendor who fit the bill for all these parameters. After seeing what they were doing for other customers, we were convinced that outsourcing the whole problem to them was the best approach.
Initially, we worried that the journey to AI automation would be long, difficult, and expensive, but after speaking with some SmartAction customers, we recognized how easy it can be to automate one call type or process at a time. The SmartAction CX team has become an extension of our own, and we couldn’t be happier with their commitment to the process of perpetual improvement to provide the best AI experience possible for our customers. Now we’re reviewing the roadmap to identify what we should automate next.”
The experience, results, and roadmap for future call automation with SmartAction that Ross describes above is not unique to Electrolux. See Gartner-vetted Reviews>>
Hear about the 4 ways other organizations – like AAA, MGM Resorts, Hyundai, and Office Depot – are leveraging the power of conversational AI through cloud-based virtual agents. Watch Webinar>>