I once burned through $500,000 of my own money on a product that went exactly nowhere. It was a social fitness app, and I was convinced I had it all figured out. We spent months building the most technically complex features you can imagine. We had an AI-powered trainer that analyzed your form, a predictive meal planner, and a feature that would generate workout music based on your heart rate. It was a masterpiece of engineering. And it was a complete and utter flop.
We launched, and the silence was deafening. All that beautiful code, all that “innovation,” and nobody cared. I was burned out, broke, and my confidence was shot. I had fallen into the classic trap of the tech founder: I was in love with the technology, not the customer.
It took me years to recover from that failure, financially and emotionally. But it taught me a lesson that has since been worth millions to me, both in my own companies and in the 200+ startups I’ve invested in. The lesson is this: not all features are created equal. And if you don’t understand the difference, you’re just rolling the dice.
My secret weapon for not repeating that mistake is a simple framework from the 1980s called the Kano Model. It has nothing to do with AI, but it’s the most powerful tool I’ve found for building AI products that people actually want to use. It’s how I decide what to build, what to ignore, and what will turn users into fanatics.
The Three Flavors of Features
The Kano Model is brilliant in its simplicity. It was developed by a Japanese professor, Noriaki Kano, and it’s not some complex algorithm. It’s just a way of thinking about features based on how customers react to them. It’s about getting inside their heads.
1. Must-be Features (The Basics)
These are the absolute essentials. The table stakes. If you don’t have them, your customers will be unhappy. But if you do have them, they won’t even notice. They’re just expected.
Think about a car. It has to have wheels. If it doesn’t, you’re not going to be a happy customer. But when was the last time you bought a car and thought, “Wow, this is amazing, it has wheels!”? You just expect it. That’s a Must-be feature.
In software, this is stuff like a login page that works, or a button that does what it says it will do. It’s the boring, unsexy stuff that has to be there for the product to be functional.
2. Performance Features (The More, The Better)
These are the features where more is better. There’s a linear relationship between how much you provide and how satisfied your customers are. The more you invest here, the happier they get.
Back to the car analogy. Fuel efficiency is a classic Performance feature. A car that gets 30 miles per gallon is good. A car that gets 40 is better. A car that gets 50 is fantastic. The more you give, the more people like it.
In AI, this is often about accuracy or speed. A translation app that’s 95% accurate is good. One that’s 99% accurate is much better. A model that generates an image in 10 seconds is good. One that does it in 2 seconds is great.
3. Delighters (The Magic)
These are the unexpected, delightful features that make a customer say “Wow!” They’re the things they didn’t even know they wanted, but once they have them, they can’t live without them. If you don’t have them, nobody complains because nobody expected them. But when you do, you create evangelists.
Remember the first time you used a backup camera in a car? You’d been parking your whole life without one. You didn’t need it. But the first time that little screen lit up and showed you exactly what was behind you, it felt like magic. That’s a Delighter.
In AI, this is where the real magic happens. It’s the feature that makes your product feel intelligent and special. It’s the surprise that creates a deep emotional connection with your users.
The Kano Model in the Wild: My Experience
This all sounds great in theory, but how does it actually work in practice? Let me give you some examples from my own career.
When we were building RemoteTeam (which was later acquired by Gusto), we were creating a platform to help companies manage their remote employees. We had a million ideas for features.
When we were building RemoteTeam—which was later acquired by Gusto—we were creating a platform to help companies manage their remote employees. We had a million ideas for features. Our must-bes were obvious, if unglamorous: accurate payroll processing and compliance with international labor laws. If we messed that up, we were done. Nobody would have cared about our cool AI features if we were paying their employees the wrong amount. That was the foundation.
For performance, a key metric was the time it took to run payroll. If it took a customer two hours, that was acceptable. But if we could get it down to ten minutes? That was a huge win. We invested heavily in making that process as fast and efficient as possible, and every minute we shaved off made our product stickier.
But the real magic, the feature that got people talking, was a delighter. We built a tool that automatically calculated and suggested the optimal time for team meetings across different time zones. Nobody was asking for it. It wasn’t on any competitor’s feature list. But when we shipped it, our users went crazy for it. It solved a huge, annoying headache they didn’t even realize could be solved. It was a small thing, but it showed we understood their world on a deeper level.
I see this pattern over and over again with the companies I invest in. I look for founders who have a deep understanding of their customers’ needs and can map them to these three categories. When I talk to a founder and they can’t tell me what their must-haves are, that’s a huge red flag. It tells me they’re probably in love with their tech, not their customer.
I passed on a company once that had built an incredible AI for generating personalized children’s stories. The tech was amazing. But they had no way for parents to actually buy the books. They were so focused on the delighter (the AI) that they had completely ignored the must-be (a functioning e-commerce experience). They ran out of money six months later.
On the other hand, one of the reasons I invested in Scale AI was because Alex and the team had a crystal-clear understanding of this. They knew that for their customers (AI developers), the must-be was high-quality, accurate data labeling. The performance feature was the speed and scale at which they could deliver that data. And the delighters were the tools and APIs that made it incredibly easy for developers to integrate that data into their workflows. They nailed the basics, then scaled the performance, and then added the magic.
Your Guide to Using the Kano Model
So, how do you actually put this to work? It’s not rocket science. It’s a four-step process that’s more about discipline than genius.
Step 1: Brainstorm Your Features
Get your team together and list out every single feature you can possibly think of for your product. Don’t filter anything at this stage. Just get it all down on paper.
Step 2: Survey Your Users
This is the most important step. You need to talk to your users. For each feature on your list, you’re going to ask them two simple questions:
- Functional question: “How would you feel if you had this feature?”
- Dysfunctional question: “How would you feel if you did not have this feature?”
For each question, they can choose one of five answers: I like it, I expect it, I’m neutral, I can live with it, I dislike it.
Step 3: Analyze the Results
Based on the answers to those two questions, you can categorize each feature. There’s a simple table you can use to do this. For example, if a user says they expect a feature (functional) and would dislike not having it (dysfunctional), that’s a clear Must-be.
If they say they would like having it and would dislike not having it, that’s a Performance feature. And if they say they would like having it but would be neutral if they didn’t, that’s a Delighter.
Step 4: Build Your Roadmap
Now you have your features sorted into the three Kano buckets. This is your roadmap. You absolutely must build all the Must-bes. No excuses. These are your foundation. Then, you need to make some choices. You can’t do everything. I recommend picking one or two Performance features where you can be the best in the market. And then, you need to find one or two Delighters that will make your users fall in love with you. That’s it. That’s your strategy.
Don’t Fall for the AI Hype
It’s so easy to get caught up in the hype of AI. Every day there’s a new model, a new technique, a new “breakthrough.” And the temptation is to chase that shiny new thing and build your product around it.
That’s a recipe for disaster. That’s how you end up with a technically brilliant product that nobody wants.
The Kano Model is the antidote to that. It forces you to start with the customer. It forces you to think about their needs, their expectations, and their desires. It grounds you in reality.
So, let’s be real. Most AI products are just tech demos looking for a problem. Don’t be that founder. Don’t fall in love with your algorithm. Fall in love with your customer. The Kano Model isn’t a magic formula, but it’s the best compass I’ve found to keep you pointed in the right direction. Stop chasing the hype and start building for delight.
Frequently Asked Questions
How has this view evolved over time?
My thinking on most topics has changed significantly over the years. Early in my career, I held many conventional views that experience proved wrong. I try to update my beliefs when the evidence changes.
What experience informs this perspective?
This perspective comes from over a decade of building companies in Silicon Valley, two successful exits (RemoteTeam to Gusto, MovieLaLa to Gfycat), and investing in 200+ startups including Anthropic, OpenAI, and Scale AI. I write about what I've lived.
How can I apply this thinking to my own situation?
Start by identifying the core principle behind the opinion, not the specific example. Then ask yourself: does this principle apply to my context? If yes, test it in a small, low-risk way before going all in.