< Home
January 3, 2020 • 10 min

I wasted $40k on a fantastic startup idea

When good ideas make bad business

You have a mind-shattering headache. You’re standing in the aisle of your local CVS, massaging your temples while scanning the shelves for something—anything—to make make the pain stop.

What do you reach for? Tylenol? Advil? Aleve?

Most people, I imagine, grab whatever’s cheapest, or closest, or whatever they always use. But if you’re scrupulous enough to ask Google for the best painkiller, here’s how your friendly neighborhood tech behemoth would answer:

Most effective painkillers

Oh thanks Google that’s just all of them.

If you’re among the 77% of Americans that Google their health problems, insipid answers like this won’t surprise you. But we should be surprised, because researchers carry out tens of thousands of clinical trials every year. And hundreds of clinical trials have examined the effectiveness of painkillers. So why can’t I Google those results?

And so in the year of our lord 2017 I had a Brilliant Startup Idea: use a structured database of clinical trials to provide simple, practical answers to common medical questions.

As a proof-of-concept I tried this by hand: I made a spreadsheet with every OTC painkiller trial I could find and used R to run a network meta-analysis, the gold standard of evidence-based medicine.

The results were pretty interesting, and exactly the kind of thing I was looking for back in the sad sterile aisles of CVS:

Most effective painkillers

Results from my first meta-analysis (157 trials, 4,400 people)

A wave of exhiliration washed over me. Here was a problem that

  1. Was interesting
  2. Could help people
  3. I knew how to solve

A perfect bullseye. After a few hours searching domains I came up with a name for my project: GlacierMD.

Over the next nine months I would quit my job, write over 200,000 lines of code, hire five contractors, create a Delaware C-Corp, add four doctors to my advisory board, and demo GlacierMD for twelve Bay Area medical practices. I would spend $40K of my own savings buying clinical trials and paying contractors to enter said trials into the GlacierMD database.

On July 2, 2018, GlacierMD powered the world’s largest depression meta-analysis, using data from 846 trials, beating Cipriani’s previous record of 522.

GlacierMD&rsquo;s evidence graph for depression treatments.

GlacierMD’s evidence graph for depression treatments.

Choirs of angels sang in my ears. Here I was, living the Silicon Valley dream: making the world a better place through technology.

Two weeks later GlacierMD was dead.


“That’s an awesome idea,” said Carl. “It sounds like something worth working on.”


Carl was my boss. We worked at a startup that leveraged autonomous blockchains to transfer money from naïve investors to slightly less naïve twenty-somethings. There are worse gigs.

And here was Carl telling me that my startup idea would bring such benefit to humanity that I simply had to quit, his roadmap be damned. I nodded knowingly, feeling the weight of this responsibility resting on my proud shoulders.

“Thanks Carl,” I said. “I’ll try to mention you when I accept my Nobel.”

I quit two weeks later and started coding at a blistering pace. I drew all sorts of inscrutable diagrams with dry-erase pens on my parents’ windows. I hired a motley crew of Egyptian contractors to start entering clinical trials into my database. I commissioned a logo, registered my domain, and started obsessing over color schemes.

Data extraction portal for GlacierMD

Data extraction portal for GlacierMD

When I finally finished the MVP I showed it to the head of product at the company I’d just left. I watched him as he watched my demo, waiting for his eyes to melt with the glory of it all. Instead he just sorta shrugged.

“Lots of people make medical claims on the internet,” he said. “Why should I trust yours?”

I started babbling about network meta-analyses, statistical power, and p-values, but he cut me off.

“Yeah okay that’s great but nobody cares about this math crap. You need doctors.”

Goddamnit he was right. If nobody could be bothered with the math, then I was no better than Gwyneth Paltrow hawking vagina eggs. To build trust I needed to get endorsements from trustworthy people.

So I called up some friends, some buddies, some friends-of-friends. “Would you like to be an advisor for my cutting-edge health-tech startup?” I’d ask, while eating Dominos in my parents’ laundry room. I’d give them 1% of this extermely valuable, high-growth startup and in exchange I could plaster their faces all over my website.

Four of these doctors agreed. This is called making deals ladies and gentlemen and I was like the lovechild of Warren Buffet and Dr. Oz.

Things are going great. My friends and family all tell me they love the site. Even some strangers on the internet love it. “I know right,” I tell them. “So how much would you pay for this?”

“Hahahahahahah,” they say in unison. “Good one!”

I forgot that the the first law of consumer tech is nobody pays for consumer tech. But no problemo, I say to myself. This is why Eric Schmidt invented ads. I’ll just plaster a few banners on GlacierMD and bing bang boom I’ll be seasteading with Peter Thiel before Burning Man.

But then I look at WebMD’s 10-Qs and start to spiral. Turns out the world’s biggest health website makes about $0.50/year per user. That is…not enough money to bootstrap GlacierMD. I’m pouring money into my rent, into my Egyptian contractors, into AWS—I need some cash soon.

What I need are people willing to pay for this thing. What about doctors? Doctors have money, right? Maybe doctors, or practices, or whatever—someone in the medical industry—maybe they would shell out some cash for my on-demand meta-analyses.

So I listened to a few podcasts and became a sales expert. I started cold calling people using scripts from the internet and tried to convince them to sit through a GlacierMD demo.

In the meantime I receive some worrying messages from my Egyptian contractors.

“I think it’s time to talk about a raise,” one of them says.

“I feel that I have become exceptional at my job,” says another. “Please consider a raise or I will stop working.”

“Please increase my pay,” says the third, including helpful screenshots demonstrating how to give said raise through the Upwork website.

Are my contractors unionizing? I wonder. I glance obliquely at my shrinking bank account statement, grit my teeth, and approve the raises. At this rate I’ll hit zero in a matter of weeks.

But my sales calls start paying off. Miraculously I find some doctors that are willing to talk to me. So I borrow my parents’ car and drive out to the burbs to meet a doctor I’ll call Susan.

Susan has a small practice in downtown Redwood City, a Silicon Valley town that looks 3-D printed from the Google Image results for main street.

Susan is a bit chatty (she’s a psychiatrist) but eventually I demo GlacierMD. I show her how you can filter studies based on the demographic data of the patient, how you can get treatment recommendations based on a preferred side effect profile, how you can generate a dose-response curve. She oohs and aahs at all the right points. By the end of the interview she’s practically drooling.

Comparing the efficacy of two antidepressants over time

Comparing the efficacy of two antidepressants over time

Hook, line, and sinker I think to myself. I’m already contemplating what color Away bags would look best in the back of my Cybertruck when Susan interrupts my train of thought.

“What a fun project!” she says enthusiastically.

Something in her tone makes me pause. “Uh, yeah,” I say. “So what would you imagine a product like this—one that could change the very practice of medicine—how much would you pay for such a service?”

“Oh, uh—hmmmm,” she said. “I don’t know if we can spare the budget here, to be honest. It’s very fun…but I’m not sure if our practice can justify this cost.”

If you read enough sales books most of them tell you that when people say your product is too expensive what they really mean is your product isn’t valuable enough. Susan acted like I was offering her Nirvana as a Service so the conversation has taken quite a wild turn.

“So you don’t think this product is useful?”

“Oh sure! I mean, I think in many cases I’ll just prescribe what I normally do, since I’m comfortable with it. But you know it’s possible that sometimes I’ll prescribe something different, based on your metastudies.”

“And that isn’t worth something? Prescribing better treatments?”

“Hmmmm,” she said, picking at her fingernails. “Not directly. Of course I always have the best interests of my patients in mind, but, you know, it’s not like they’ll pay more if I prescribe Lexapro instead of Zoloft. They won’t come back more often or refer more friends. So I’d sorta just be, like, donating this money if I paid you for this thing, right?”

I had literally nothing to say to that. It had been a bit of a working assumption of mine over the past few weeks that if you could improve the health of the patients then, you know, the doctors or the hospitals or whatever would pay for that. There was this giant thing called healthcare right, and its main purpose is improving health—trillions of dollars are spent trying to do this. So if I built a thing that improves health someone should pay me, right?

I said goodbye to Susan and tried to cheer myself up. I had ten more meetings with doctors all over the Bay Area—surely not all of them were ruthless capitalists like Susan. Maybe they would see the the towering genius of GlacierMD and shell out some cash.

But in fact everyone gave me some version of Susan’s answer. “We just can’t justify the cost,” a pediatrician told me. “I’m not sure it’s in the budget,” said a primary care physician. “It’s awesome,” said a hospitalist. “You should try to sell this!” Ugh.


So in July 2018, nine months and $40K after starting GlacierMD, I shut it down. I fired my contractors, archived the database, and shut down the servers. GlacierMD was dead.

Make something people want. It’s Y-Combinator’s motto and a maxim of aspiring internet entrepreneurs. The idea is that if you build something truly awesome, you’ll figure out a way to make some money off of it.

So I built something people wanted. Consumers wanted it, doctors wanted it, I wanted it. Where did I go wrong?

Occassionally I like to disconnect from the IV drip of internet pseudoknowledge and learn stuff from books. I know, it’s weird—maybe even a bit hipster. But recently I read Wharton’s introductory marketing textbook, Strategic Marketing Management. The very first chapter has this to say:

“To succeed, an offering must create value for all entities involved in the exchange—target customers, the company, and its collaborators.”

- Strategic Marketing Management

All stakeholders. You can’t just create value for the user: that’s a charity. You also can’t just create value for your company: that’s a scam. Your goal is to set up some kind of positive-sum exchange, where everyone benefits, including you. A business plan, according to this textbook, starts with this simple question: how will you create value for yourself and the company?

What they teach you in business school.

What they teach you in business school.

I winced audibly when I read this. How much time I could’ve saved! If I’d articulated at the beginning how I expected to extract value from GlacierMD, maybe I would’ve researched the economics of an ad-based model, or I would’ve validated that doctors were willing to pay, or hospitals, or insurance companies.

A few months after shuttering GlacierMD and returning to corporate life my buddy pitched me a new startup idea.

“It’s called Doppelganger,” he said. “It’s super simple—you upload a selfie to the database, and then it uses AI or whatver to instantly find everyone in the database who—”

“Looks like you,” I finished for him.

“Exactly,” he said, grinning ear to ear. “How awesome would that be? You should build it!”

I mean, I dunno, it sounds like something fun to do at parties. In a narrow sense, it’s something I want, but there’s no way in hell I’m going to devote any time to this. Doppelganger has created value for the customer but not for the company.

“Call me when you have a business plan,” I said, lacing up my Allbirds and riding my Lime scooter into the sunset.


Get an email when I post stuff: