Your average systems integrator is still designing digital experiences from left to right. They’re thinking about the happy path. Customer starts at point A, goes to point B, goes to point C, buys something, and then leaves.
That’s great. But that’s not where the value is.
The easy data, the kind of data that makes for flawless demos, lives in that happy path. The hard data, the kind that actually matters, lies in everything that went wrong:
- The missed self-service attempts.
- The abandoned carts.
- The escalated calls.
- The conversations where customers said, “This didn’t work.”
If you want AI to deliver real business value, not just incremental efficiency, you need to design right to left. Start with failure—those contact center conversations you already know about—and create hybrid digital experiences that seamlessly blend digital-only, AI-assisted, and human-supported experiences.
The Blind Spot in “AI-Enabled CX”
“AI projects don’t scale.” You’ve heard it for the last year. The signals are everywhere:
- AI is bolted onto platforms, not embedded in them.
- Self-service adoption remains low. Customers still call anyway.
- Expensive platforms sit underutilized.
- AI pilots stall before reaching production.
Here’s the thing: these aren’t technology problems. The reason 80% of those AI projects don’t scaleⁱ, and 95% of organizations are getting zero returnsⁱⁱ, comes down to two things: data and design.
Failure is Where the Most Valuable Data Lives
Over the past 20 years, companies have been making incremental investments in data platforms tied to critical business operations. Every year or two, technology changes. They stand up a new technology. Then another.
But the old technologies never go away.
Data fragments across silos, trapped in bolted-on technologies that don’t share, don’t communicate, and don’t allow for meaningful extraction. The irony is that many of those technologies were added to keep customers on the happy path, to reduce friction.
There’s value in resolving customer needs, but there’s far more value in understanding where the experience breaks:
- Where customers abandon journeys.
- Where self-service fails to resolve issues.
- Where context is lost across channels.
- Where agents step in to recover the experience.
This is failure intelligence, and it’s one of the most underutilized assets in enterprise CX.
Unlike traditional analytics, failure intelligence is grounded in actual customer behavior under real conditions. It shows you not what customers are supposed to do, but what they actually do when things don’t work.
That makes it the most reliable starting point for AI design.
Designing Right to Left
AI is just another technology being bolted on, while the data or the systems beneath it become an afterthought. Too many organizations are building AI on top of experiences that were never designed to handle failure in the first place.
They’re automating the happy path and leaving the friction untouched.
That’s why even well-funded AI programs struggle to scale. The data hasn’t been unified, and the underlying experience hasn’t been built for it.
Designing right to left flips the model. Instead of asking, “How should this journey work?” you start with, “Why did it fail?” Then you redesign the experience backwards:
- Start with contact center insights: Identify the highest-volume failure points—repeat contacts, handoffs, unresolved queries.
- Connect those insights to digital journeys: Map failures back to specific moments in web, app, or self-service experiences.
- Fix the experience at the source: Redesign journeys to eliminate friction rather than just push it downstream.
- Embed AI where it matters: Apply conversational AI, automation, and predictive routing where failures actually occur.
This is the difference between adding AI to an experience and rebuilding the experience to work with AI.
The New Realities of CX Design
Too many companies ask how to deploy AI in their CX. Here’s the better question: how do you ensure AI has access to the right data and is integrated into the right enterprise applications?
Experience design itself must change in four fundamental ways:
1. From journeys to systems:
Experiences are no longer linear. They bounce around, crossing channels, platforms, and data ecosystems, often overlapping with other experiences.
AI only works when those systems are connected, when it has access to the same context, data, and workflows as human agents. Otherwise, you’re just adding friction to broken journeys.
2. From pilot to production
Up to 85% of AI initiatives fail to move beyond the pilot stage.ⁱⁱⁱ Not because of a lack of ambition, but from operational complexity and data readiness.
Designing from failure helps prioritize the use cases that matter and ensures they’re built to scale from day one. It’s the only way to prevent fragmented fixes from collapsing under their own weight.
3. From silos to cooperation
Executing on AI requires cooperation across legacy silos, including digital teams and contact center teams. These silos are exactly why organizations struggle to scale pilots to production.
While pilots can be done within a leader’s sphere of control, production requires a comprehensive view of input from multiple stakeholders.
4. From reactive to proactive optimization
Experience design isn’t a one-time effort. It’s a continuous loop: capture failure signals, diagnose root causes, redesign journeys, and deploy AI to prevent recurrence.
Instead of a loop where every fix creates a new problem down the road, this design loop compounds—every improvement reduces friction, cost, and customer effort.
Your Customer Experience Failures
When you start designing right to left, the impact goes beyond data and CX metrics. You unlock:
- Lower cost-to-serve through reduced contact volume.
- Higher platform ROI by activating underused capabilities.
- Faster AI adoption by focusing on high-impact use cases.
- Better customer outcomes through consistent, connected experiences.
More importantly, you move from reacting to customer problems to preventing them entirely. In CX environments where one problem can spark a viral complaint on social media, a journey that works—consistently—becomes your strongest marketing strategy.
Start Where it Hurts Most
The next time you think about redesigning your customer experience—especially to deploy AI—don’t start with the journey map, and don’t start with the next technology.
Start with the complaints.
Start with the conversations your customers are already having.
Start with the data and what it tells you about where your opportunities lie.
Because the fastest path to better experiences doesn’t begin with perfection, it begins with failure.
Resources
ⁱ “Why AI Projects Fail,” RAND, Apr 10, 2025
ⁱⁱ “The GenAI Divide: STATE OF AI IN BUSINESS 2025,” MIT, July, 2025
ⁱⁱⁱ “The GenAI Divide: STATE OF AI IN BUSINESS 2025,” MIT, July, 2025
