6 min read

How to Flail at Doing a Startup, Part 1: The Story So Far

I have five months off before I go full time. I chose this break so I could work on projects, because I had never gotten around to it during school.

The original plan was to build an MVP a month like Pieter Levels.

But I’ve been working on Quickapply for the last month and a half, and I’m not sure when I’ll stop. The MVP lets you apply to 40 software engineering internships with one application.

In this post, I share what’s been going on with Quickapply the last month and a half. This is best as a read for those curious about the early stages of starting a project.

Origin story

I came up with the concept last December during winter break. To test demand, I posted on my school’s CS Facebook page:

Screen Shot 2019-11-22 at 12.51.49 PM.png

I had 4 customers paying $49 to apply to 100 internships.

I also asked a lawyer to see if it was legal here.

At this point, I had to decide whether anyone wanted it. To be honest, I had no idea. I could explain the data either way: enough people may or may not want it.

I was then supposed to build it over the summer while I was interning, but I didn’t. I procrastinated. I started building October 1st, 2019.

MVP development

I wanted to launch in three weeks, but it took six.

I didn’t write down a spec, which was the first mistake. I had some notes describing the functionality, but I never looked at them again. As a result, I had crazy scope creep.

I kept a loose reign on time, and to be honest, I wasted dozens of hours. For one thing, I rewrote code and tweaked copy over and over. I couldn’t help myself.

For another, I didn’t tear myself away from tasks once I realized how time consuming they would be. Splitting the Quickapply application into three parts. Automating Facebook’s internship application. Setting up rishi@quickapply.io.

Each of those things cost me an engineering day or more.

I also spent three days trying to bypass Recaptcha so I could scale, which was a mistake. Wtf, Rishi. You don’t have any customers, you don’t need to scale.

I came to my senses when I didn’t finish by end of October and then cut scope. I knew an MVP should be light, but I didn’t realize the amount of discipline needed to achieve that.

If I was building another MVP, I would write a product spec and whittle it down. Question every feature if it’s necessary. Then, I’d write out engineering tasks and estimates. For each task, I’d ask, what’s the fastest way to do this? Is there a shortcut?

So different from programming at work.

Product market fit

“If you have to ask whether you have Product/Market Fit, the answer is simple: you don’t.” Eric Ries, author of The Lean Startup

I do have to ask, so maybe I don’t have product market fit, and I just haven’t admitted that to myself yet.

Here’s my data, you tell me:

4 customers paid $49 each to apply to 100 places in January, after I got 500 views (estimate) on my school’s CS Facebook page.

2 customers used it for free to apply to 40 places this week, after I got 300 views (estimate) on r/cscareers. 50 people opened the form, and these two completed the form.

Screen Shot 2019-11-20 at 10.03.42 AM.png

One of those customers made my day and said:

Screen Shot 2019-11-22 at 2.47.04 PM.png

The pricing suggestions hurt me but let’s ignore that for now LOL

These words suggest I at least have product user fit. But product market fit?

These questions remain:

  • Are people willing to pay and how much?
  • Will enough people pay to make it worthwhile?
  • How much does it cost to get a user?

A friend explained if I hustle to 50 user interviews and 50 users, I can probably tell if the market exists or not. So I’ll try that.

But the million dollar question: Should I kill the project, or nah?

Internal conflict over whether to kill the project

A part of me wants to build something, anything, right this instant. I’m OK with small business ideas. I’d rather just get started. Looking for the biggest ideas just invites procrastination.

Another part of me wants to work on only the best ideas. Does it have the characteristics of a monopoly? Would it get into YC? Could it be the next unicorn?

One of my favorite indie hackers, Patrick McKenzie, wrote this:

If I got a do-over on e.g. running my own businesses for a decade, I’d probably skip the relatively unambitious first two (which I did just because I knew I could probably do them) and proceed directly to trying things outside of my comfort zone. (I still have an enormous love in my heart for the aesthetics of small software businesses, but given a do-over, could have done much more interesting and impactful small software businesses earlier than I actually did.)

For context, his first product was straight up a bingo card creator. But he learned a ton about business (long tail SEO, anyone?), and got to financial independence with it.

But regardless. Both parts of me might agree Quickapply is bad business. To check if a business is any good, you can use YC Partner Kevin Hale’s criteria, or Thiel’s monopoly criteria. To check if a market is any good, you can use The Personal MBA’s market criteria.

Here are Quickapply’s undesirable characteristics:

  • The problem it’s trying to solve is infrequent. Students usually apply to jobs once a year, and working professionals even less. Customers can’t use the product again until the next year.
  • There’s a ton of upfront work to solve the problem. Each application I add to the index, can take as little as ten minutes, but often a half an hour. This work would have to be redone every year. Huge pain.
  • There are no monopoly characteristics, especially not economies of scale. You need humans to fill out Recaptchas, and so the cost to serve a customer is not pennies, but at least $7.50.

I keep flip flopping on whether to continue with Quickapply. When I first drafted this post last week, I intended it to be a post mortem. But there’s still more I can learn from this project, so I’ll run it for a while longer.

A couple weeks of user interviews and traction tests, at least. Then, I’ll have to decide again whether to continue.

Business model

Unit economics

Cost to serve a customer:

Assuming I can’t bypass Recaptchas, a human needs to man the script as it runs, and fill out Recaptchas. Assuming 1-2 applications an hour, and a minimum wage of $15 an hour, cost to serve would be $7.50-$15.

Cost to acquire a customer:

I need to test traction channels to figure out this number. More on this in the marketing section.

Traction channels with high CAC are out of the question, like search engine marketing. I read the CAC for SEM >= $30.

Willingness to pay:

How much are customers willing to pay? I’ve now upped the price to $9. Hopefully, there’s a good response.

Is it possible that people are willing to pay up to $50, if I up the number of companies from 40 to 100? I’ve decided to ask customers this on the signup form.

TAM (Total Addressable Market):

TAM doesn’t seem like a useful number to think about to me.

I’m more wondering, can the traction channels I use deliver a lot of customers. Seems more concrete to think about.

Idk, let me know what you think there.

Potential pivots:

Upsell customers on resume critiques and mock interviews, or even coaching and courses. I’ve always known the money is in these information products. But I have to ask myself: am I after profits? Then why don’t I pursue information products? Or am I after making software products?

Screen Shot 2019-11-22 at 3.11.27 PM.png

Have the tool become so popular companies pay to be added to the index: This is a whole ’nother can of worms, lol


Dunno much about marketing. I read this book, Traction: How Any Startup Can Achieve Explosive Customer Growth, which has been a great help. It has a simple framework to pursue traction called Bullseye + a thorough rundown of 19 traction channels. I’m testing the two following channels to start.

Offline ads: Flyers on student CS lab bulletin boards. I’m having a friend place flyers at UCSD. I’ll check if there’s a promising response. Will use Google Analytics to see which requests are coming from SD.

Targeting blogs: I’ve found some blog posts on getting internships that rank well on Google search results. I’ll reach out to the authors and ask them to link Quickapply. I’ll have to pay them, rip.

Content marketing could be the move for this area, but it’s a long term play, and I’m not sure if I’m in it for the long term.


Holy moly, thanks for reading.