I once blew half a million dollars on an AI analytics platform that told me absolutely nothing.
There, I said it. It was one of the most expensive lessons of my career, and it came dressed in a beautiful dashboard full of charts that went up and to the right. The problem? Those charts had zero connection to our actual revenue. We were celebrating vanity metrics while the business was slowly bleeding out.
Most founders I talk to are making the same mistake. They’re so mesmerized by the promise of “AI-powered insights” that they forget to ask the most important question: what business problem are we actually trying to solve? They get sold on fancy tech and end up with a data ghost town. A pretty interface with no one home.
The Siren Song of the Useless Dashboard
For three years, I was lost in the wilderness of AI analytics. At RemoteTeam, we were obsessed with data. We tracked everything: user engagement, feature adoption, click-through rates, you name it. We had dashboards for our dashboards. And for a while, it felt like progress. We’d have meetings, point at a chart, and say, “Look! Engagement is up 15%!” Everyone would nod and feel smart.
But our growth was flat. Stagnant. We were a classic case of being data-rich and information-poor. The tools we were using were great at telling us what happened, but they were useless at telling us why it happened, or what was going to happen next. It was all rearview mirror stuff. That’s not analytics; that’s reporting. And reporting doesn’t get you to your next funding round.
I remember one board meeting where I proudly presented a deck showing a 30% increase in daily active users. One of my investors, a sharp woman who’d seen it all, just looked at me and asked, “And how many of those new users are on a paid plan?” The answer was a rounding error. We were celebrating freebie-seekers who were churning out after a week. The data was telling a story, but it was a work of fiction.
The $500k Disaster That Changed Everything
The breaking point came when we signed a massive contract with a big-name AI analytics vendor. They promised us the world: predictive lead scoring, automated churn prediction, the whole nine yards. We spent months integrating their system. The price tag was a cool $500,000.
The result? A system so complex, so opaque, that it was practically unusable. It spit out “predictions” that were consistently wrong. It told us a user was about to churn, and they’d upgrade the next day. It flagged a lead as “hot,” and they’d be a college student writing a research paper. It was a black box of nonsense.
After six months of frustration, I pulled the plug. It was a painful decision, but it was the right one. We had to go back to first principles. We had to stop looking for a magic box and start thinking like scientists.
Cracking the Code: From Rearview Mirror to Predictive Engine
We fired our expensive vendor and assembled a small, scrappy team. An engineer, a data scientist, and me. Our goal was simple: build a predictive analytics engine that was directly tied to revenue. No more vanity metrics. Every single insight had to answer the question, “How does this help us grow the business?”
We started with a massive data dump. We pulled over 10 million data points from every system we had: our CRM, our product database, our support tickets, everything. It was messy. It was chaotic. But it was real.
Instead of looking for correlations, we started looking for causation. We built a simple model. What sequence of actions did a user take before they upgraded to a paid plan? We found a pattern. It wasn’t about how many times they logged in. It was about whether they used three specific features in their first week.
Bingo.
That was our “aha” moment. We had found a leading indicator of revenue. Not a lagging one.
Here’s the framework we developed. It’s not sexy, but it works:
Start with a Business Question, Not Data. Don’t ask, “What can the data tell us?” Ask, “What’s the most important problem we need to solve?” For us, it was converting free users to paid. Everything flowed from that.
Identify Your “Magic Moment.” What’s the one action a user takes that shows they’re getting real value from your product? For Facebook, it was connecting with 7 friends in 10 days. For us, it was using those three specific features. Find your magic moment and you’ve found the core of your growth engine.
Build a Simple Predictive Model. You don’t need a team of PhDs from Google. We built our first model in a Google Sheet. It was a simple lead score based on a handful of variables. Is the user’s company in our target industry? Did they invite a team member? Did they hit the magic moment? We weighted these actions and came up with a score.
Test and Iterate. We started feeding our sales team the leads with the highest scores. The results were immediate. Our conversion rate from free to paid tripled in two months. We went from a $500k data disaster to a 300% growth engine. We kept refining the model, adding new data points, and throwing out the ones that didn’t work. It was a constant process of experimentation.
Stop Buying AI, Start Building It
The biggest lie the AI industry tells you is that you need to buy their expensive, complicated tools to succeed. You don’t. You need to understand your business and your customers. You need to be relentlessly focused on solving a real problem.
I’ve invested in over 200 companies, including some of the biggest names in AI like Anthropic and OpenAI. The founders who succeed are the ones who treat AI as a tool, not a religion. They’re not afraid to get their hands dirty. They build simple, focused solutions that drive real business results.
So, the next time a salesperson tries to sell you a fancy AI dashboard, ask them one question: “How is this going to help me find my next paying customer?” If they can’t give you a straight answer, show them the door. Your business depends on it.
Frequently Asked Questions
Do all experts agree with this view?
No, and that's fine. The best ideas in business are often contrarian. I share my perspective based on my experience and data, but I encourage you to seek out opposing viewpoints and form your own conclusions.
What's the most common pushback you get on this?
People often push back by citing exceptions or edge cases. And they're usually right that exceptions exist. But building a strategy around exceptions rather than patterns is a losing game for most founders.
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.