AI Can Save Bad CX, but Not How You Expect
There's been a lot of buzz recently about the idea of artificial intelligence customers. Meaning: AI agents that act and transact on the customer's behalf. At Forrester Research, we refer to these as digital doubles.
Companies are beginning to think about what this transition will mean for the future of customer service, commerce, advertising, you name it. Websites and content will need to be optimized for machine navigation and consumption. We will need to develop protocols for AI agents to engage with businesses so they are not forced down current human pathways, clogging up all our queues.
This future is not here yet. It might not be here for some time. But companies are starting to prepare.
I recently read about Virgin Atlantic testing OpenAI's Operator for internal user experience optimization. They're experimenting to understand how AI systems navigate their websites ahead of future AI traffic. At the same time, they're using Operator as a human customer proxy to pinpoint UX improvements for today's users.
Now that got me thinking. Where might companies use this technology to improve customer service experiences for us human customers while we wait to hand things off to our AI overlords?
Today's Contact Centers Struggle to Maximize Self-Service Containment
Contact centers know there are friction points in the digital experience, but they lack the actionable insights to diagnose them effectively. Analyzing conversation transcripts can be powerful, but never perfect. Too often, companies deploy chatbots to support customers navigating flawed self-service workflows.
Just a free tip from me to you: If your customers need a chatbot to help them use your website, your website is too dang complicated!
That's why using an AI agent to stress-test self-service workflows for common call drivers could be a game changer! It could simulate customer navigation and pinpoint pain points, such as unclear instructions, broken paths, or missing knowledge. Even if the experience is technically sound, the AI agent could provide recommendations to reduce effort and minimize self-service abandonment.
Even companies that have achieved solid self-service containment will struggle with the dreaded long tail. Uncommon issues, tricky exceptions, multi-intent compound scenarios, cross-channel navigation are all kisses of death for successful self-service. Most teams would just consider the long tail uncontainable (if they had even identified these issues in the first place). But an AI agent could test these edge cases to see where things break.
I'm getting ahead of myself, but imagine a world where conversation intelligence vendors reach upstream by offering stress-testing agents. Once top call drivers are identified, AI agents could be automatically deployed to find pain points and recommend fixes.
Sure, sure. AI will take over the world, and we will no longer need contact center agents. I hear you.
But, you know, just in case that takes a while to play out…
Companies should apply the same stress-testing principle to have AI agents identify bottlenecks in workflows for front-line reps. Employee-facing UX gets far too little investment and attention. Let an AI agent follow your onboarding documentation or process guidance to flag areas where the process is inefficient, unclear, or downright broken. Have the AI agent complete your reps' most common workflows to identify automation opportunities. That task that takes three minutes every time it comes up (approximately 3 million times a year)? Turns out it's super-automatable.
Every week, I talk to contact center leaders eager to deploy technology to support front-line reps during calls. But we're making the same mistake we make with self-service chatbots: layering AI-shaped band-aids over bad UX instead of fixing the real issues driving those contacts.
The truth is, real cost savings don't come from managing or containing contact drivers more efficiently; they come from eliminating those drivers altogether. Solve the underlying problems that make customers reach out in the first place, and you'll eliminate those contacts, not just redirect them.
Christina McAllister is a senior analyst at Forrester Research, covering customer service and contact center technology, strategy, and operations.