I once burned through $50,000 building a feature not a single person used. Zero. It was early in my career at my first startup, and I was so sure we were building the next big thing. We did all the “right” things. We ran focus groups. We sent out surveys. The data looked solid. Then we launched. And… crickets.
That failure was a painful, expensive, but incredibly valuable lesson: building what customers say they want is a fantastic way to burn money. People are surprisingly bad at knowing what they need. They’ll give you a laundry list of features, but they can’t articulate the deep-down problem they’re trying to solve. That’s the “job” they’re hiring your product to do. If you don’t get that, you’re just guessing.
This is the core of the Jobs-to-be-Done (JTBD) framework, and it completely changed how I think about building companies. But here's the thing: JTBD alone is a great map, but AI is the rocket fuel.
What the Hell is Jobs-to-be-Done?
Clayton Christensen’s milkshake story is the classic way to explain JTBD. A fast-food joint wanted to sell more milkshakes. They asked people what they wanted: thicker, more flavors, lower price. They tried it all. Sales didn’t budge.
A researcher came in and just watched people. He saw a ton of milkshakes were sold in the morning to people who were alone in their cars. He started talking to them. It turned out they had a long, boring commute. They weren't buying a milkshake for the taste. They were hiring it to do a job: keep them occupied and full during their drive. The milkshake wasn't competing with other milkshakes. It was competing with bananas, donuts, and bagels.
That insight changed everything. They made the shakes easier to drink with one hand and sales shot up. That’s JTBD. It’s about the “why,” not the “what.” After my own $50k feature disaster, I read everything I could find on product development. Christensen's work hit me like a ton of bricks. I realized we’d been so obsessed with building a better drill that we never asked if anyone actually wanted a hole in their wall.
Why Your User Research is Probably Lying to You
My early mistake is why I’m so skeptical of traditional user research. We asked, and people answered. The problem is, what people say and what they do are worlds apart. Focus groups are a joke. People want to be helpful, so they tell you what they think you want to hear. Surveys are even worse. You get a lot of data, but very little real insight.
At my second company, MovieLaLa, we were building a social network for movie fans. We asked our users what they wanted, and they all said more social features. So we built them. Friending, messaging, groups. Nobody used them. We were on the verge of killing the whole project. Then we tried something different. We interviewed our most active users. But instead of asking what they wanted, we asked them to walk us through the last time they used the app. Turns out, they weren't there to connect with friends. They were there to find something new to watch. That single insight led to a complete pivot. We were later acquired by Gfycat.
AI: The Secret Weapon for JTBD
So if you can’t trust what people say, how do you find the real job? This is where AI becomes your secret weapon. AI can chew through mountains of data—support tickets, reviews, social media chatter, in-app behavior—and find patterns a human could never spot. It’s a goldmine of messy, unstructured, but honest data.
At RemoteTeam, my last company, we pointed an AI at thousands of our customer support tickets. We used natural language processing (NLP) to pull out the most common themes and sentiment analysis to see how people were feeling. We found a huge number of our customers were struggling with managing their remote teams. They were using our product for a job we hadn't even designed it for. That discovery led to a whole new set of features built specifically for that job. It was a massive success, and Gusto eventually acquired us.
A No-BS Guide to Combining JTBD and AI
You don’t need a data science PhD to do this. Here’s a simple, five-step guide:
- Get specific. What exactly are you trying to learn? Who are you learning about? Don't be vague.
- Collect everything. Don’t just run surveys. Talk to your customers. Read their support tickets. Look at their in-app behavior. More data is better.
- Let the AI do the heavy lifting. Use off-the-shelf tools or build your own to find the patterns. Look for the recurring themes.
- Write your “job stories.” A job story is a simple sentence: “When [situation], I want to [motivation], so I can [expected outcome].”
- Validate with real people. Take your job stories back to your customers. Do they resonate? This is the most important step.
The Real Future of Product Management
I’m convinced that the combination of JTBD and AI is the future of building great products. We’ve been operating on guesswork for too long. With AI, we can finally understand our customers on a deep, fundamental level. We can build products they don’t just use, but love.
This isn’t just about building better companies. It’s about solving real problems for real people. That’s a job worth doing.
It won’t be easy. It’s a new way of thinking and requires a new set of skills. But I promise you, it’s worth it. The future isn’t about more features. It’s about more understanding. And that’s a future I’m excited to build.
Frequently Asked Questions
How do I measure success with this approach?
Pick one or two metrics that directly tie to your goal and track them weekly. Vanity metrics like page views or follower counts rarely matter. Focus on metrics that reflect real engagement or revenue impact.
Do I need technical skills to combine jobs-to-be-done and ai for breakthrough insights?
Not necessarily. While technical understanding helps, the most important skills are clear thinking and the ability to break problems into smaller pieces. Many successful founders I've invested in started with zero technical background and either learned enough to be dangerous or found the right technical partner.
What are the most common mistakes when combining jobs-to-be-done and ai for breakthrough insights?
The biggest mistake I see is overcomplicating things early on. Start with the simplest version that works, get real feedback, and iterate from there. Another common trap is copying what worked for someone else without understanding the context behind their decisions.
What tools do I need to get started?
Start with the basics. You don't need expensive software or fancy tools. A spreadsheet, a note-taking app, and direct access to your customers will get you further than any enterprise platform. Add tools only when you hit a specific bottleneck.