Bad CX Is a feature, not a bug—why AI vendors are designed to frustrate your customers
When deflection-driven pricing creates customer frustration by design, not by accident
Customer experience has reached a breaking point. Forrester's 2025 Global Customer Experience Index shows CX quality at an all-time low in North America, with 25% of U.S. brands declining for the second consecutive year while only 7% improved. Meanwhile, PwC research shows that 59% of customers will walk away after several bad experiences, and 17% after just one.
Bad customer experiences are a feature, not a bug—the predictable result of how AI vendors get paid.
Gladly
Here's the uncomfortable truth nobody wants to say out loud: Those frustrating bot loops that trap customers for 20 minutes aren't programming failures. They're not an oversight. It’s not even poor design.
Bad customer experiences are a feature, not a bug—the predictable result of how AI vendors get paid.
When customers rage-quit after arguing with a chatbot that refuses to connect them to a human, that's not broken AI. That's a business model working exactly as designed.
The economic incentive behind customer frustration
Most AI vendors only profit when their bot "resolves" an issue without human involvement. Every handoff to an agent costs them money. The math is simple and brutal: if a vendor charges for "deflection" or "containment," then every customer who reaches a human represents lost revenue.
This creates a perverse incentive structure where vendors program AI to avoid escalations at all costs. The bot keeps customers trapped in endless loops, forces them to repeat information, suggests irrelevant articles, or simply provides generic responses until customers give up. The vendor marks this as a "successful deflection." The customer marks it as the last time they'll buy from you.
Think about it: Would you design a car where the airbag manufacturer loses money every time it deploys? Would you trust a smoke detector company that only profits when there's no fire? That's exactly how most customer service AI works today. The worse your customer's experience, the better the vendor's margins.
Simple chatbots don't want to highlight when their product doesn't meet expectations; they're incentivized to show off a higher deflection rate because it increases their perceived value. When a chatbot "deflects" a customer by providing a generic response that doesn't solve their problem, the vendor gets paid even though the customer remains frustrated.
How deflection metrics corrupt customer service
The obsession with deflection has created a vocabulary of failure masquerading as success. Companies celebrate "containment rates" as if preventing customers from reaching help is victory. They measure "avoidance rates" as if dodging customer needs is an achievement. They track "self-service adoption" while customers frantically search for phone numbers.
These metrics tell a story, but it's the wrong story. They measure efficiency for the company, not effectiveness for the customer.
Consider a typical deflection-first interaction: A customer contacts support about a damaged product. The AI asks for an order number, then asks again. It suggests articles about product care when the customer clearly states the item arrived broken. It offers to check shipping status when the problem is physical damage. After 15 minutes of circular conversation, the bot proudly provides a returns portal link—something the customer could have found in 30 seconds.
The vendor marks this as a win. The customer marks this as the last time they'll buy from you.
While deflection rate shows you the share of interactions your customer service team avoided due to your bot's intervention, a high deflection rate doesn't necessarily mean all customer issues were resolved. This is the critical flaw in deflection-only thinking.
AI CX platform comparison matrix of solutions and key capabilities
The hidden costs of misaligned incentives
When vendors only profit from blocking human contact, the damage compounds far beyond individual frustrations. Research shows that customers will pay up to 16% more for better experience, yet 32% will abandon a brand they love after just one bad experience.
But there's a cost CFOs rarely calculate: agent burnout. When AI only deflects, human agents become the dumping ground for the angriest, most frustrated customers. Every conversation starts with someone who's already upset, who's already wasted time, who's already considering leaving. Agents receive no context from the AI interaction—they can't see what was tried, what failed, or why the customer finally broke through to human support.
This creates a doom loop. Agents burn out from handling only the worst interactions. Service quality drops. Customer frustration increases. More customers fight to reach humans. The cycle accelerates until the entire support organization breaks down.
"Customer experience continues to erode worldwide, reflecting a concerning multiyear downward trend and a shift in sentiment from positive to neutral. While the changes may seem subtle, they are significant and cannot be ignored."
Pete Jacques
Principal Analyst, Forrester
Why resolution-only metrics fail everyone
The fatal flaw in resolution-only thinking is that it assumes every customer problem fits neatly into a bot-solvable box. Reality is messier. Sometimes customers need empathy. Sometimes they need expertise. Sometimes they need creative problem-solving that no AI can provide.
The companies winning with AI understand something critical: Great service means knowing when AI should resolve AND when it should assist. They've stopped seeing human involvement as failure and started seeing it as intelligent design.
Resolutions happen when AI completely solves an issue. Password reset? Order status? Business hours? These are perfect for AI. No escalation needed, customer satisfied, case closed.
Assists happen when AI makes humans more effective. The AI gathers context, identifies the real issue, pulls relevant history, and seamlessly transitions to an expert who already knows the full story. No repetition. No starting over. No frustrated explanation of what just happened with the bot.
Both create value. Both deserve recognition. Both should be part of how AI success gets measured.
What success actually looks like
Outdoor brand KÜHL demonstrates this balanced approach. They achieved 59% AI resolution while seeing a 120% increase in revenue per conversation. How? Because their AI knows when a customer asking about waterproof jackets needs a quick specification (resolved) versus when they need expertise on arctic expeditions (assisted). The customer gets the right level of help, the agent gets context for meaningful consultation, and the sale happens because trust was maintained.
Breeze Airways shows what's possible when AI is designed for outcomes, not deflection. They've achieved 71% of conversations assisted by AI—not deflected, assisted. Through intelligent prioritization and automation, they've decreased handle times by 45%, with AI fully resolving 37% of conversations while maintaining high satisfaction scores.
The difference is philosophy. While deflection-first models trap customers to avoid costs, these companies use AI to enhance every interaction—whether that ends in full automation or intelligent human handoff.
The leadership test that reveals everything
Want to know if your AI vendor's incentives align with great customer experience? Ask this single question:
"What happens to your revenue when AI hands off to a human agent?"
If they say "we lose money" or "we minimize those," you've identified the source of your customer frustration. If they dodge the question, talk about "containment rates," or pivot to discussing "deflection optimization," you're looking at a vendor whose business model depends on keeping customers away from help.
This question cuts through the sales pitch and reveals true incentives. It's the difference between a partner invested in your success and a vendor profiting from your customers' frustration.
The honest answer you want to hear is: "We value both resolutions and assists because both create customer value." Any other answer means their success comes at your customers' expense.
The real cost of CX decline
As Pete Jacques, principal analyst at Forrester, notes: "Customer experience continues to erode worldwide, reflecting a concerning multiyear downward trend and a shift in sentiment from positive to neutral. While the changes may seem subtle, they are significant and cannot be ignored."
Companies that view customer service as a value center rather than a cost center achieve 3.5 times greater revenue growth.
Forrester
This isn't just about individual bad experiences. Companies that view customer service as a value center rather than a cost center achieve 3.5 times greater revenue growth. When AI vendors design systems that prioritize their margins over customer outcomes, they're not just creating bad experiences—they're destroying long-term business value.
Research consistently shows that 86% of buyers are willing to pay more for a better customer experience. Yet when AI systems are designed to minimize human contact rather than maximize customer success, companies lose both immediate revenue and future loyalty.
Building the future of CX
The path forward isn't about choosing between AI and humans. It's about designing systems where both thrive—where AI handles what it does best (speed, consistency, data processing) while humans handle what they do best (empathy, creativity, complex problem-solving).
Progressive companies share three characteristics:
First, they measure complete outcomes, not just containment. They track whether customers achieved their goals, not whether they stayed trapped in the bot.
Second, they value seamless handoffs as much as full automation. They invest in context preservation, intelligent routing, and ensuring no customer ever repeats themselves.
Third, they choose vendors whose business models align with customer success. They partner with companies that profit from satisfaction, not frustration.
The competitive advantage of aligned incentives
Companies fixing this now will dominate their industries. While competitors frustrate customers with deflection-first AI, they'll build loyalty through intelligent assistance. While others leak revenue through poor experience, they'll grow through word-of-mouth advocacy and premium pricing power.
The disparity between the customer experience that brands intend to deliver and what customers actually experience is widening. Organizations that bridge this gap by aligning vendor incentives with customer outcomes will capture disproportionate market share.
Bad customer experiences aren't inevitable. They're choices. And the first choice is picking an AI partner whose success depends on your customers' success, not their frustration.
The AI revolution in customer service isn't about replacing humans—it's about amplifying human capability. It's about knowing when to automate and when to accelerate. It's about measuring what matters: not how many customers you deflected, but how many you delighted.
When you understand that bad CX is a feature built into most AI vendor business models, you can finally make the choice to find partners who profit from your success instead of your customers' surrender. The question isn't whether AI will transform customer service—it already has. The question is whether that transformation will create better experiences or worse ones.
The answer depends entirely on how your AI vendor gets paid.
AI CX platform comparison matrix of solutions and key capabilities
Recommended Blogs
AI CX platform comparison matrix
Make an informed AI CX platform decision with our comprehensive side-by-side comparison of leading solutions and their key capabilities.
CX buyer’s checklist
Key features to compare when choosing an AI-powered CX platform.
Test your CX maturity in 5 minutes
See where your customer experience stands today — across AI, routing, service, and personalization.