The Hairless Kittens Who Taught Me Everything About User Onboarding (And What Your Digital Product Can Learn from This)

So remember me? The cat lady who finds UX lessons everywhere? Well, this week I learned that my 13-year-old twin daughters are apparently living, breathing edge cases that can crash any system not designed for the beautiful chaos of real life. Picture this: routine blood work for both girls at one of those big lab facilities. Same doctor's orders, same appointment time, same mom dragging them there on a Saturday morning. Should be simple, right? I walk up to this fancy digital check-in kiosk. You know the type - all touchscreen and "Please enter your information to streamline your visit!" First twin steps up: "Enter patient's last name and date of birth." Easy. Morley, January 15th. System: "Welcome! Please take a seat." Second twin's turn: "Enter patient's last name and date of birth." Same info: Morley, January 15th. System makes these confused beeping sounds. Screen: "Patient already checked in. Please see reception." The guy at the front desk looks at my girls, looks at his screen, then looks back at my girls like they're some kind of medical anomaly. "The system thinks they're the same person," he says, clicking around frantically. "It can't process two patients with identical last names and birthdates." I'm watching this poor guy spend 15 minutes trying to manually override a system that had apparently never considered the existence of twins. He had to call IT support. For twins. In 2025. My daughters just stood there like, "Mom, why is this taking so long?" Meanwhile I'm thinking about all the times I've seen this exact scenario in software I've designed. Twins aren't exactly rare. There are about 110,000 sets born in the US every year. But somebody designed this system assuming no two patients would ever share a name and birthday. Classic edge case oversight. The receptionist ended up creating a workaround by adding "(Twin 1)" and "(Twin 2)" to their names in the system. My daughters didn't need to change anything about themselves. The technology needed to get smarter. And that error message? "Patient already checked in" told us absolutely nothing useful. Something like "Multiple patients found with this information. Please see reception for assistance" would have saved everyone confusion. I guarantee this system worked perfectly when the development team tested it with made-up patient data like "John Smith, June 1st" and "Jane Doe, July 2nd." But real families are messier than test data. Whether you're building a SaaS platform, mobile app, or booking system, you're probably designing for the 80% of normal use cases. But that other 20%, the twins, the people with one name, the customers who want to do things slightly differently, they're not bugs in your system. They're features of being human. So here's what I ask myself now when I'm designing anything: What happens when users enter unexpected but totally valid data? Have I tested with real people doing real things, or just perfect test scenarios? Do my error messages actually help users figure out what to do next? Both girls got their blood drawn (they're fine, thanks for asking). The receptionist was actually really sweet about the whole thing and handled it with humor. And I got another great story about why good UX design means planning for the beautiful, chaotic reality of how people actually live. Next week I might tell you about the time my soccer kid taught me everything about user preferences by completely customizing her cleats, or maybe what happened when I let my Irish dancer redesign our candle website navigation. I've got 25+ years of designing digital products, but my best insights still come from watching real people navigate real life with all its wonderful complications.

Let's be honest: implementing AI is hard. Really hard. Over the past few years, I've been fascinated by the stories of companies integrating AI into their products and internal workflows. The excitement is always palpable at the beginning, the possibilities seem endless, the potential transformative. But then reality sets in. Even with fantastic technical teams and significant investment, organizations struggle to translate AI capabilities into solutions that users actually adopt and value. It's not surprising that research shows over 80% of AI implementations fail to deliver on their promises. What separates the successful 20% from the rest? After studying numerous AI implementation cases, I've noticed it's rarely the sophistication of the technology. Instead, it's the thoughtfulness of the user experience design surrounding that technology. Here are five critical strategies that can dramatically increase your odds of successful AI implementation. 1. Start with User Needs, Not AI Capabilities One of the most common patterns in failed AI projects is companies becoming entranced by AI's capabilities and looking for places to apply it, rather than starting with genuine user needs and determining whether AI is actually the right solution. I recently read about a SaaS company that was determined to implement a sophisticated AI-powered recommendation engine because their main competitor had just launched one. When they finally stopped to ask what problem they were trying to solve for users, the team struggled to articulate a clear answer. The Better Approach: Begin with a deep understanding of your users' actual pain points and workflows. Map their journey, identify moments of friction, and look for opportunities where AI could remove obstacles rather than create new ones. For companies facing this challenge, taking a step back to conduct focused user research often reveals that what users actually need isn't what was initially assumed. For instance, users might want faster access to relevant information when making complex decisions rather than automated recommendations—leading to an intelligent search system that surfaces contextual information at key decision points. 2. Design for Appropriate Trust, Not Maximum Automation Another common pitfall is companies designing for maximum automation rather than appropriate trust. They focus on what the AI can do rather than what it should do given the context and stakes of the decision. In healthcare technology, for instance, AI systems designed to automatically prioritize patient cases without clinician input often create significant resistance from medical staff who feel their expertise is being sidelined, even when the technology works as intended. The Better Approach: Map decisions along two axes: consequence and complexity. For high-consequence or highly complex decisions, design AI to augment human judgment rather than replace it. For lower-stakes, more routine decisions, greater automation might be appropriate. The most successful healthcare AI implementations present case prioritization as suggestions with clear reasoning, allowing clinicians to quickly review and adjust if needed. This collaborative approach addresses concerns about professional judgment while still delivering the efficiency benefits of AI. 3. Create Transparent Interfaces That Build Understanding Organizations implementing AI-powered systems often struggle with the "black box" problem. Users don't understand how decisions are being made, leading to mistrust and resistance. In financial services, for example, AI-powered fraud detection systems that simply flag transactions as "suspicious" based on AI analysis, with no explanation, often generate frustration from both customers and service representatives. The Better Approach: Design interfaces that make AI decision-making transparent in user-friendly ways. This doesn't mean exposing complex algorithms, but rather translating AI reasoning into understandable terms that build user confidence. Effective fraud detection interfaces create simple visualizations showing which factors contributed to the suspicion score, using plain language explanations like "unusual location" or "pattern differs from typical spending." This transparency helps reduce complaints and helps service representatives resolve cases more quickly. 4. Build Learning Loops Into Every AI Experience When users provide feedback to AI systems but don't see visible improvement, they quickly lose faith in the technology. This is a particularly common issue in content recommendation and analytics platforms. For instance, when users repeatedly dismiss irrelevant recommendations or insights but the system continues offering similar ones, it creates the impression that the AI isn't actually "learning" or responding to feedback. The Better Approach: Design explicit feedback mechanisms into AI experiences and make learning visible to users. When people see that their input improves the system, they become invested in its success. Successful implementations include simple feedback systems where users can indicate whether insights are useful, along with a quick reason why or why not. Adding a "Getting smarter" notification that appears when the system incorporates feedback to improve personalization can dramatically boost long-term engagement with AI features. 5. Implement AI Gradually, With Clear Metrics for Success Organizations eager to embrace AI often attempt to transform everything at once, new recommendation engines, chatbots, dynamic pricing, and personalized experiences all simultaneously. This approach typically overwhelms both internal teams and end users. The Better Approach: Implement AI in phases, starting with focused use cases that deliver clear value. Establish specific metrics for success beyond technical performance, including user adoption, satisfaction, and business impact. Companies that take a measured approach, starting with focused implementations like product recommendations in specific categories, create room for careful measurement and refinement before expanding to other areas. Each new AI capability should be evaluated against specific adoption and business metrics before moving to the next phase. The Human Element Is Your Competitive Advantage The most successful AI implementations share a common trait: they treat the human experience as a primary design consideration, not an afterthought. As AI capabilities become more widely available, the technical differentiators between competing solutions will diminish. The real competitive advantage will lie in how thoughtfully organizations integrate these capabilities into human workflows and experiences. This is precisely why UX leadership is becoming increasingly critical in AI implementation. It's not enough to have data scientists and engineers creating powerful algorithms. You need experience designers who understand human psychology, behavior, and needs to translate those technical capabilities into solutions people actually want to use. Putting It All Together: A Framework for Success When approaching AI implementation, consider using this simple framework that integrates the five strategies: Identify: Map user journeys to find high-value opportunity areas where AI could solve real problems Evaluate: Assess each opportunity based on potential value, technical feasibility, and appropriate level of automation Design: Create transparent interfaces with clear feedback mechanisms, focusing on building appropriate trust Implement: Roll out capabilities in focused phases with clear success metrics Learn: Collect user feedback, measure against success metrics, and continuously improve This user-centered approach to AI implementation consistently delivers higher adoption rates, greater user satisfaction, and ultimately, better return on AI investment. The Path Forward As AI capabilities continue to evolve, the organizations that succeed won't simply be those with the most advanced technology, they'll be the ones that create thoughtful, human-centered experiences that build trust and deliver genuine value. If you're embarking on an AI implementation journey, I encourage you to put user experience at the center of your strategy from day one. It might seem like it slows you down initially, but it dramatically increases your odds of being in the successful 20% rather than the failing 80%. And that's a much better return on your AI investment. Ready to discuss how user-centered design can help your organization successfully implement AI? Let's chat.

We've all been there. You're using a new AI-powered feature, and suddenly it does something unexpected. Maybe it's suggesting products you'd never buy, displaying information that seems completely off-base, or making a recommendation that leaves you wondering, "How on earth did it come up with that?" And just like that, trust is broken. As someone who's spent years in UX design and who follows developments in AI implementation closely, I've become fascinated by one critical question: Why do users trust some AI systems but reject others? The Trust Gap in AI Adoption Here's a sobering reality: over 80% of AI implementations fail to deliver their expected value. Not because the technology doesn't work, but because people don't use it. And why don't they use it? It usually comes down to trust. Trust isn't just a nice-to-have in AI user experience, it's the entire foundation. Without it, even the most sophisticated AI investment becomes an expensive experiment that never delivers. The Three Pillars of Trustworthy AI Experience Based on extensive research and observation of successful AI implementations, I've identified a framework for building AI experiences that users actually trust and embrace. It's built on three fundamental pillars: 1. Transparency Without Overwhelming Users need to understand what the AI is doing without drowning in technical details. This means: Explaining recommendations in human terms Making clear when AI is being used versus traditional functionality Providing appropriate confidence indicators Offering access to more information for those who want it Consider financial services, where AI-based investment recommendations often go ignored by users. A simple "Why am I seeing this?" feature that explains each recommendation in plain English, connecting it to specific financial goals, can dramatically increase adoption. 2. Control at Critical Moments People need to feel they're working with AI, not being controlled by it. This means designing for: User confirmation at key decision points Easy ways to guide, adjust, or override AI suggestions Clear feedback mechanisms Progressive levels of automation as trust builds In healthcare applications, for example, successful AI symptom assessment tools allow users to adjust the importance of each symptom they report, ensuring they feel heard and respected. This control mechanism can increase both user satisfaction and diagnostic accuracy. 3. Continuous Improvement Through Feedback Trust grows when users see the system getting better based on their input. This means creating: Simple feedback loops within the user journey Clear signals when feedback has been incorporated Personalization that visibly improves over time Acknowledgment when the AI gets something wrong Marketing analytics platforms that implement one-click feedback systems for AI-generated insights can create a virtuous cycle where the system visibly improves for each user, driving deeper engagement. Practical Implementation: Start Small, Build Trust If you're incorporating AI into your product experience, you might be tempted to go big with flashy, highly automated features. My advice after studying dozens of successful AI adoptions? Start small. Begin with low-risk, high-value use cases where AI can clearly enhance the user experience without taking over completely. For example: Smart defaults that save time but can be easily changed Subtle suggestions that guide without dictating Background enhancements that make existing features work better In retail, for instance, companies often find more success starting with AI-powered search that simply reorders results based on browsing history rather than implementing fully automated personalization across the entire shopping experience. This creates immediate value while allowing users to discover the benefits of personalization in a comfortable, low-stakes context. The Human Element Remains Essential The most successful AI experiences maintain a decidedly human element. They acknowledge limitations, handle errors gracefully, and maintain the brand's personality and voice. When designing conversational AI for customer service, the most effective systems recognize when a customer is getting frustrated and smoothly offer to connect them with a human agent. This "escape hatch" often increases overall satisfaction with the AI system because users never feel trapped. The Future of AI User Experience As AI capabilities continue to evolve at breathtaking speed, the human side of the equation becomes even more critical. The organizations that succeed won't simply be those with the most advanced algorithms, they'll be the ones that create thoughtful, human-centered experiences that build trust and deliver genuine value. The most powerful AI implementations don't feel like interacting with technology at all, they feel like having a particularly insightful colleague who understands exactly what you need and helps you accomplish it with minimal friction. That's the gold standard for AI UX: technology that feels less like a tool and more like a trusted partner. Building Trust from Day One If you're embarking on an AI implementation journey, make user trust a central consideration from the very beginning. Ask questions like: Where in the user journey would AI add the most value with the least disruption? What level of explanation would users need to feel comfortable with AI-driven decisions? How can we give users meaningful control without overwhelming them with options? What feedback mechanisms will help the system visibly improve over time? These considerations might seem secondary to the technical implementation, but they're actually the primary determinants of whether your AI investment will succeed or join the 80% that fail to deliver value. Ready to Design AI Experiences Users Actually Trust? If you're navigating the complex landscape of AI implementation and want to ensure your investment translates into user adoption and business value, I'd love to chat. I'm passionate about helping organizations create AI experiences their users actually trust and embrace.

Learn how to build and implement effective design systems that scale. Discover practical strategies for improving development efficiency, ensuring consistency, and driving adoption. Expert guide by Michelle Morley based on enterprise implementation experience.

Explore how fractional UX leadership can transform your product development. Learn when and how to leverage experienced design leadership without the full-time commitment. Expert insights from Michelle Morley on implementing successful UX strategies.