The 9X Problem: Why Being Better Isn't Enough

Rdio was the music streaming service that critics loved. TIME named it one of the "50 Best Websites" in 2014. Entertainment Weekly gave it a straight A. The interface was cleaner than Spotify's, the social features were better thought through, and the music discovery engine was, by most reviewers' assessments, superior. The people who used Rdio evangelized it with the intensity of early Apple adopters.

On November 16, 2015, Rdio filed for bankruptcy. At the time of its death, it had a fraction of the market. Spotify had twenty million paying subscribers worldwide and growing fast. Pandora bought the remains for seventy-five million dollars — not for the product, not for the users, but for the patents and the engineering team.

The product was better. The market didn't care.

If you've ever built something you knew was better than the alternative and watched the market shrug, you're not imagining the problem. You're experiencing a gap between how you evaluate your product and how your customer's brain evaluates the cost of switching to it. That gap has a specific size, a specific mechanism, and a specific name, and it explains more startup failures than bad products, bad timing, or bad marketing combined.

The 9X Mismatch

John Gourville, a professor at Harvard Business School, published an article in 2006 in Harvard Business Review called "Eager Sellers and Stony Buyers" that quantified what most founders experience but can't explain.

Creators overvalue their innovations by about three times. This isn't arrogance — it's structural. The endowment effect inflates the value of what you own. The IKEA effect inflates the value of what you built. The curse of knowledge makes it impossible to simulate what the product looks like to someone encountering it fresh. Three separate biases, all pointing in the same direction: your product is worth more to you than it is to anyone else.

Meanwhile, customers overvalue what they already have by about three times. Loss aversion, status quo bias, and the sheer weight of familiarity all compound — the same loss aversion that kept the Concorde flying long after the math said to stop. The brain treats "what I'm currently doing" as the default, and defaults carry the weight of a high-precision prediction: this is working, don't change it. Every day the customer uses their current approach without disaster, the brain accumulates more evidence that the current approach is good enough.

Multiply those together. Three times three. The result is a nine-to-one mismatch between what you believe your product is worth and what the customer's brain computes when faced with the decision to switch.

Being twice as good isn't enough. Being three times as good isn't enough. To overcome the 9X gap, your product needs to be roughly ten times better — or you need to fundamentally change the computation.

The Curse of the Creator

In 1990, a Stanford graduate student named Elizabeth Newton ran a simple experiment that captures the 9X problem at its most visceral.

She assigned participants to one of two roles: tappers or listeners. Tappers received a list of well-known songs — "Happy Birthday," "The Star-Spangled Banner" — and tapped the rhythm on a table. Listeners tried to identify the song.

Before the experiment, Newton asked tappers to predict how often listeners would recognize the song. Tappers predicted fifty percent. The actual recognition rate was 2.5 percent. Forty times worse than expected.

The tappers were astonished. To them, the song was obvious — they could hear the melody in their heads while they tapped. They couldn't fathom how the listener heard only a random series of taps. The knowledge inside their own heads was so vivid, so present, so clearly connected to the song, that they were neurologically incapable of simulating the experience of someone who didn't have that knowledge.

This is what happens every time a founder demos their product. You see the vision. You hear the melody. The customer hears taps on a table. The gap between your experience and theirs isn't a communication failure. It's a computational one — your prediction engine has been trained on thousands of hours of context the customer doesn't have, and it can't untrain itself.

Google learned this the hard way with Wave. In 2009, they launched what was supposed to be the future of communication — real-time collaborative documents and messages, combined into a single interface. The engineering was brilliant. The hype was enormous. A hundred thousand invitations went out.

Then people used it. The live-typing feature — where collaborators could see every keystroke as you composed a message — made users so self-conscious they stopped writing. The interface combined too many concepts into a single view for anyone who hadn't spent months with the product to parse. Fewer than a million people actively used it. Google killed it in August 2010, barely a year after launch.

The Wave team could hear the melody. They knew where every feature lived, how every interaction connected, what every icon meant. They had spent years tapping. The users heard noise.

The Status Quo Is a Decision

Rdio's failure wasn't about features. It was about what the product asked the customer to do. Rdio launched with a paid-only model. To use it, you had to stop whatever you were currently doing for music and start paying for something new. Every element of that sentence is a friction point the brain evaluates as a cost: stop a current behavior (loss), start a new one (effort), pay money (pain of paying), all in exchange for a benefit the brain can only simulate hypothetically.

Spotify launched with a free tier. You didn't have to stop anything. You didn't have to pay anything. You could use Spotify alongside whatever you were already doing, with zero switching cost. The free tier wasn't a business model concession. It was a 9X solution — it eliminated the computation that kills adoption by removing the loss, the effort, and the payment from the initial evaluation.

Once inside the free tier, Spotify's prediction-error engine did the rest. Every good song recommendation was a small surprise — the same reward prediction error that drives all dopamine-mediated learning — better than expected. Every playlist that matched your mood was a prediction confirmed, building the precision of the model, making the service feel increasingly indispensable. By the time the upgrade prompt appeared, the brain's model of "what I'm currently doing" had already been rewritten. Spotify wasn't an alternative anymore. It was the status quo. The 9X gap now worked in Spotify's favor.

This is the pattern that separates products that overcome the gap from products that die in it. You don't argue your way past inertia. You eliminate the switching cost entirely at the point of entry, and then let the product rewrite the customer's prediction model from the inside.

The First-Time User Test

The most reliable way to see the 9X gap is to watch it happen in real time.

Stewart Butterfield, before Slack reached its explosive growth, discovered what he called the "magic number." His team analyzed user behavior and found that teams who exchanged two thousand messages almost never left — ninety-three percent retention. Below that threshold, retention collapsed. The product wasn't different for the two groups. The experience of the product was different, because the users who reached two thousand messages had generated enough prediction errors — enough "this is useful in a way I didn't expect" moments — to rewrite their model of what a communication tool should be.

Butterfield didn't find this number by asking customers what they thought of Slack. He found it by watching what they did. The gap between what customers say and what customers do is the 9X problem in miniature: their stated evaluation runs on different hardware than their actual behavior. (This is the same gap that makes customer interviews so dangerously misleading — people will tell you they love your product while their behavior says otherwise.)