kaashif's blog

Programming, with some mathematics on the side

We sent the worst YCombinator application possible


There are many mistakes we made when halfheartedly trying to get funding for our startup. The worst was that we didn't actually have a business at all - we had no users yet due to not having regulatory approval for our financial product, and no practical plan to get that other than sending applications and hoping. The second worst was that our funding applications were really bad in many ways.

Not having a business is obviously a bigger problem than anything else, but being unable to convince investors is a big deal too, especially if you're really bad at it.

Here's a list of mistakes we made in our YCombinator application.

Preamble: How to avoid these mistakes

I watched this video: https://www.ycombinator.com/library/6t-how-to-apply-and-succeed-at-y-combinator but then we made all the mistakes and did the opposite of that advice.

I don't know what happened in my cofounder's case, but I think my excuse is that I was too focused on writing software rather than building a business. It should be as easy as building a real business, then laying out the facts so that YC can understand what you're doing.

I'll try to avoid much more commentary on how to succeed, since I didn't. Instead, let's focus on how to fail, which I do know something about.

Poor spelling and grammar

This should go without saying, but apparently not. My co-founder, a non-native speaker, wrote the application and gave it to me to proofread. I'm a native English speaker, but I was writing code on a 36 hour sleep cycle revolving around exhaustion, so I didn't spot many of the mistakes.

I spotted some of the mistakes, but this just led to unproductive arguments. For example: we had "we've done X in USA". It should be "the USA" or, more naturally, "the US". We talked about this mistake, but I hit my cofounder with the old "trust me I'm right" and brushed aside any request for explanation.

Our final application still had the mistake.

I think there's something deeper going on here - if there isn't a deep level of trust between cofounders, you can't get much done. Even if we both had the same level of English proficiency, something else would've gone wrong down the road and produced a suboptimal outcome.

This is why firms exist in the first place, it's easier for incentives to be aligned and for people to trust each other within a firm. A firm without trust may as well not exist.

Stretching the definition of user

An example: how many active users did we have? None, since we didn't have regulatory approval yet. But we put down 114 users, since hundreds of people had registered their interest and thus..."used" our website, right? I mean that's obvious bullshittery, don't do that.

The worst part is that we started to treat "interest" as a key performance indicator (KPI). We got excited and thought we were accomplishing something when thousands of people came to our site and signed up.

We hadn't done anything! We shouldn't have believed our own lies about user numbers. There's only one kind of user that matters, and that's one that's giving you money, because revenue and profit are the only real KPIs.

Embellishing progress

We had contacted several regulators about approval of our product. We had something that seemed to be workable. That was good. But we hadn't received approval from anyone yet. Indeed, we hadn't received replies from any regulators yet at the time of sending the application.

We said we were in "talks" with regulators, and wrote a long list, including regulators we hadn't contacted yet but we "surely" would. We of course didn't indicate that we hadn't contacted all regulators in our list.

We also claimed to have a minimal viable product (MVP), and that was true from a technical perspective. But from a regulatory perspective (the important one), it was unclear that our methods of sending money would be allowed. So it wasn't viable at all. We still claimed to have an MVP. This was obviously false, since we didn't have any paying users.

Delusional projections

An obsession with projections many years out is always a bad sign, and we didn't have that. We had to come up with some numbers for the "How much could you make?" question for the YC application, and there were some facts (the size of the market, average fees) but the "possible" market share we picked was completely made up.

We also made up a future growth rate (modest, we thought) and "demonstrated" that we'd be making billions within 5 years.

We knew the projections were unrealistic, but then something funny happened: we started to believe them just a bit. The act of projecting that we'd be billionaires made us think it'd be true, to a tiny extent. It's almost literally like writing "I'll be a billionaire" into a cell in Excel - meaningless in reality, but it may have some psychological effect.

This made us confident (for no reason) in our projected billions despite being pre-revenue. I can't help but think this kind of delusion has to hurt an application, it sure was obvious in how we phrased our predictions.

Complete humour incompatibility

This is important. I can laugh at something with my friends and say "that's fucking hilarious, can you believe that?" My cofounder and I never did that.

This made the "tell us something surprising or amusing" question very difficult.

I made a suggestion - it's literally cheaper to fly to some countries with a suitcase full of cash than it is to send money electronically. And it'll even be faster! And there are flights on weekends, but sometimes no bank transfers! Crazy, right?

We couldn't find a single thing we both found funny, and ended up writing that we'd both started and quit (different) degree programs.

Not being able to agree on something interesting or funny was in itself interesting to me. I wasn't able to run a company with someone I hadn't shared a real laugh with, despite not thinking that was particularly important beforehand.

A good lesson for me, I think.


Being friends with your cofounder is far, far, far more important than literally anything else. Doesn't matter what skills either of you have, whether they complement each other, whether you have the right backgrounds, etc.

The company can't survive if the cofounders don't like and trust each other. That means the founders should've known each other for a while and ideally have built something non-trivial together, so that they can trust in each other's abilities.

I think YC's co-founder matching is very stupid if you have any other choices. It should be a last desperate resort when you've asked everyone you know, and everyone they know too. It probably won't work and in the worst case, you only realise it's not working after months of work and slowly building distrust.