In part one, we looked at five of the ten most common automation mistakes organizations make when using Conversational AI. Today, we’re here with part two to help you implement and automate right the first time, and maximize your time to value and ROI.
Mistake 6: Wasting Effort on Needless Complexity
You’ve no doubt got a wide range of processes that could be automated and you’re already imagining the kudos from above you’ll get when things are in full swing. But don’t get carried away too soon.
Find the best processes to automate by looking at the volume of conversations, cost of conversations, and ease of automation.
Sometimes a very simple process like authentication can be automated easily at scale and deliver huge a increase in customer and employee satisfaction. In fact, for new projects it’s best to start with simpler processes to gather initial experience before moving into more complex areas. The simpler you start, the easier it is to monitor and optimize as you learn by doing.
Example: Small talk, while seemingly simple turns out to be a bottomless pit of complexity. It’s often one of the first topics people try to address yet delivers little to no value as customers are engaging with a virtual agent for an informational or transactional request, not chit-chat. In addition, the huge variety of greetings, responses, and slang can quickly lead to hours of unnecessary work being done to perfect something which does not help users.
Tip: Identify low-hanging fruit for automation, meaning the processes that are the most commonly used and would thus deliver the largest impact on value for customers.
Mistake 7: Treating Automation as Task Replication
As Gartner noted, “Using automation tools to copy exactly what is being done manually misses a critical benefit of automation — improving the end-to-end process to create a better customer and employee experience.” Indeed, automation isn’t just copying an existing way of doing things. It creates brand-new opportunities.
Example: You decide to replace a legacy tree-based IVR with a Conversational IVR. Instead of merely making existing processes smoother, you can find opportunities for customers to use self-service directly via CAI, such as changing reservations, making a payment, requesting a refund, etc
|Tip: Treat automating contact center processes as an opportunity to revisit established processes, because new technology might enable even better processes that deliver better UX.|
Mistake 8: Fire & Forget Launch without a Post-Launch Plan
Just because you’ve automated several processes, does not mean they’re different from any other project or IT implementation. They still require hands-on monitoring to ensure consistent and accurate performance and that everything works as expected (particularly as it handles more edge cases or new scenarios).
Example: You’re an airline that’s launched conversational IVR and self-service enabling customers to change reservations for example. When pilot strikes at a major airport throw your schedule into chaos and your daily call volume spikes to 30,000, you need to ensure your new automated system continues to work as intended. Analyze conversation paths and intents afterwards to identify new intents and flows to build so you can further optimize the customer experience.
Tip: Monitoring is key. Constantly monitor the performance of the bot from a technical and UX perspective. Link to Insights. Also, monitor for usage spikes to ensure performance.
Mistake 9: Using Vanity & Wrong Metrics
The goal of your automation project is customer and agent experience, not just moving KPI numbers one way or another. The numbers, after all, reflect the impact of the process on your business. For example, the number of conversations handled isn’t a good metric. Successful transactions are (e.g. a completed rebooking, invoices retrieved, etc.).
Customer satisfaction KPIs can also be difficult because often they aren’t rating the process itself, but their outcome is based on feelings (“what? I am not getting a free replacement. 1 out of 5 stars”). So, if your customers are now spending 25% less time on a call and resolving more issues the first time, that’s great. Even if they are not satisfied with your policy, the process is a success and that part of the experience is better.
Example: Average handling time is one of the most tracked KPIs. When implementing CAI, organizations typically see AHT increase over the long term. And yet, that’s a sign of success exactly because humans are handling primarily complex and difficult cases. Self-service and automation are handling the rest and are not there to bring the average down. Choosing the right KPIs and whether your interpretation needs to change is critical to understanding the actual impact of automation.
Tip: Identify what success looks like in advance of deployment and try to do so as a customer story, such as “I call Company X to rebook a flight, but don’t have to wait on hold and can instead verify my identity, access my reservations and change it myself with confidence.”
Mistake 10: Ignoring Employee Impact
Automation, especially large-scale, appears to present a threat to employees, something not helped by scaremongering news articles these days. Yet, statistics show it’s quite the opposite. Particularly in customer service, humans are more valuable and sought after than ever. Yet, you need to address both real and imagined concerns in advance.
Conversational AI, for example, will reduce low-value, repetitive work that agents dislike and give them more interesting and rewarding cases to handle. In addition, new roles and responsibilities will appear for conversation design, monitoring, and optimization which presents a good opportunity for existing employees to expand their skills and role.
Example: You announce a CAI project to provide chatbots on the web and conversational self-service via your new IVR. With cost-cutting rumors and a slowing economy, agents immediately fear for their jobs when they should be looking forward to handling less repetitive work and more interesting inquiries instead.
Tip: Engage employees early explaining the project scope, use cases, and benefits to employees. This will not only allay any fears but highlight new opportunities and room for growth some may be seeking.
Automation tends to encourage people to hyperfocus on processes and lose sight of the big picture which includes employee impact, interactions with other departments, an after-launch plan, and more. Automation via Conversational AI means not just improving existing processes, but new opportunities to change and improve them. Now go forth and automate wisely!