How To Make It as a
First-Time Entrepreneur

How to Make it as a First-Time Entrepreneur

Archive for the ‘ technology ’ Category


Next posts on how to acquire users for free and how to raise a Series A. Don’t miss them by subscribing via email or via twitter.

Back in 2006, I snuck out of my finance job and stood at a Midtown Manhattan Barnes and Noble wearing a full suit staring blankly at the “Computer Books” section.

Scanning through the shelf, I found “Learning HTML”, “Java in 24 hours”, “Javascript for Beginners” and other book titles of the format “crazy acronym you haven’t heard of” + “super welcoming phrase like ‘for beginners’, ‘in 24 hours’ or ‘step by step'”.

Unlike previous misguided adventures to the “Computer Books” section, I had done some research and knew that I was supposed to get the book about a “lamp”. I grabbed the closest one I could find “Apache+MySQL+PHP” (the “amp” part of “lamp”) and flipped through the first few pages. I excitedly rushed back to work. I was leaving my finance job in a year to build a tech company and I was going to learn to code.

I didn’t learn to code. I spent nights and weekends trying to teach myself. I took my programming books with me on vacation. But, despite going through all the exercises and writing a “to-do” list app and a “blog” app, I never really learned.

A year and half later (now summer of 2007), I did leave my finance job to start a tech company. But, instead of building it myself, we hired an outsourcer to build a prototype of our first big idea. We could focus on user acquisition and business development, the outsourcer would take care of the coding till we could recruit a CTO.

Nine months later, everything had gone wrong. It was clear the outsourcer wasn’t working out and, despite everything we tried, we couldn’t convince someone to join us as our CTO.

Our tech startup wasn’t going to happen unless I actually learned how to code.

So, in the beginning of 2008, I again found myself at the “Computer Books” section of that same Midtown Manhattan Barnes and Noble. I grabbed the “Learning Python” book and walked straight home.

This time, I wasn’t excited; I was terrified.

If I didn’t learn to code, we were done. I would have to crawl back into the world of finance. I’d have to tell all my friends and family that I had given up, that I had completely failed.

Three months later, not only did I finish the book, but I had re-built the prototype that our outsourcers had spent 6 months building. I was hosting it on a server I set up and we were pushing new features and iterations in hours instead of weeks. I had learned to code. 

I wasn’t ready to become a Google engineer but I could build any prototype we wanted. A few years later, we launched Yipit and we’re now a 25-person, venture-backed startup on the verge of profitability. It changed my life.

Why was this attempt to learn to code different from all the others?

Why did I learn to code? It’s simple. I had no other option.

Truly learning to code your own prototypes is incredibly hard and frustrating. I had to learn endless things including HTML/CSS, MySQL, Python/Django, Javascript, AJAX, nginx and more. I had to spend hours googling error messages praying that someone on StackOverflow had answered it and that I could understand their answer.

I found that there are two types of people that power through the frustration:

  1. Those that are really intellectually interested in learning to code. If you haven’t learned to code by now, it’s highly unlikely you’re one of them.
  2. Those that learn to code as means to an end. They don’t learn to code because it’s fun or because it’s interesting. They learn to code because they need to. They might enjoy it, almost everyone does. But, it’s different for them. They are learning to code because either their job requires it or because there’s something they need built and no one will build it for them.

So, if you’re looking to learn to code, don’t just buy a book or sign-up for a coding course.

If you really want to learn to code, you should do two things:

  1. Think of a project that you really want built and learn enough to build that project.
  2. Put yourself in a position where you have no other option other than to make sure that project gets built.

Vinicius Vacanti is co-founder and CEO of Yipit. Next posts on how to acquire users for free and how to raise a Series A. Don’t miss them by subscribing via email or via twitter.


Vinicius Vacanti is co-founder and CEO of Yipit. Next posts on how to acquire users for free and how to raise a Series A. Don’t miss them by subscribing via email or via twitter.

It was February 2010. We were noticing that a company named Groupon was taking off and a whole new industry was booming along with it.

Our idea for Yipit was simple, aggregate all these daily deals being sold by different companies and put them in one email. Plus, you could specify categories so that your emails were personalized.

The only issue was that we were working on a different product. The idea of trying yet another concept was exhausting.

So, we compromised by giving ourselves 3 days to build it.

In the first two days, we quickly built an email capture, sign-up flow to collect preferences and a script that would send people an email with the deals that matched their preference.

But, how would we get the deals in our database and categorize them correctly as “restaurant” or “concert tickets”? We would have to build a crawler to parse the deals from HTML from various sites and write a classification algorithm. Not a daily task for us.

So, we took a shortcut. Instead of building a crawler, my co-founder and I would crawl out of bed at 3 am and manually enter the deals into our database. Plus, when you’re doing it yourself, classification was easy. We did it all manually.

Within 3 days, we released Yipit and it took off. We got press and became known as the leader in the industry. Three months later we raised $1.3 million and a year later another $6 million. Today, we’re 25 people, over a million people have signed-up and we’re on a path to profitability.

The funny thing is that within a few months of our launch, several competitors emerged and they all had crawlers. But, from our users’s perspective, we were more advanced since we had categorization which was definitely no easily automated task.

We didn’t actually build a real crawler for the first 9 months and just kept scaling manually by hiring more data entry professionals. Instead, we were able to focus our resources on improving the product and user acquisition.

It’s now clear to me that not building that crawling technology early on was one of the reasons our startup succeeded.

Taking this “manual-first” approach was our secret sauce.

Many Startups Take the “Manual-First” Startup Approach

We’re not alone in our “manual-first” approach:

  • AngelList started with Nivi and Naval manually collecting startup applications and manually matching them up with potential investors. I know because Yipit was one of the first startups to use AngelList to raise funding
  • ZeroCater, a Y Combinator company, started with just a big spreadsheet trying to connect companies with restaurants that would cater
  • Groupon started with just a WordPress blog and manually sending PDFs with the first vouchers
  • Grouper, another Y Combinator company, also started with just a spreadsheet trying to match groups of people on dates

Benefits of the “Manual-First” Startup Approach

There are many benefits to taking a “manual-first” approach to some of the trickier technology challenges including:

  • Fastest way to get to the “moment of truth”. Having your potential customer evaluate your product and see if it addresses their need is the moment every founder is trying to get to and doing things manually allow you to quickly get there.
  • Easy to change your solution if it doesn’t work. There’s no code to re-write, there’s no sunk cost. You just have to change how you’re manually doing something.
  • Will really understand what to automate with tech when you’ve been manually doing it. When you’ve been manually providing the solution, you’ll know exactly where the pain points are that you should be automating.
  • Can really wow your potential customers. When you do things manually, you can try different things that really wow the customer and see which ones are worth trying to scale.
  • Customer doesn’t know how your product works behind the scenes. They won’t judge you for your manual approach because they don’t know that’s how you’re doing it. All they will care about it is that your product works.
  • Your product will “just work”. Because you’re manually providing the solution, the product will just work. When trying to implement a solution with technology, it can be very hard to make sure that it just works.
  • Helps you focus your time on the problem, not the solution. It’s very tempting to fall in love with the technology behind your solution only to painfully realize that the problem you set out to solve isn’t a real problem.

Next time you’re building a new product, I hope you’ll consider a manual-first approach to some of the trickier aspects of your solution.

Reid Hoffman, founder of LinkedIn, once said: “If you’re not embarrassed by your first release, you probably spent too much time on it.”

I also think it’s true that: “If people don’t laugh at how you first implemented your product, you probably spent too much time on it.”

Vinicius Vacanti is co-founder and CEO of Yipit. Next posts on how to acquire users for free and how to raise a Series A. Don’t miss them by subscribing via email or via twitter.

Twitter Should Allow Location Info For Each Tweet

January 6, 2009 | Comments Off on Twitter Should Allow Location Info For Each Tweet | geo, technology

Twitter Logo

At tonight’s New York Tech Meetup, I will be talking about performing data analysis based on Twitter’s revolutionary data set.  As part of my preparation work, I noticed that a number of interesting location-based services could arise from Twitter’s data set if twitter allowed each tweet to be associated with a specific latitude and longitude.

You are probably thinking that Twitter already lets you specify a location.  You are right, but that location is just a default location assigned to each user on registration.  What I am suggesting is allowing twitter users to submit updates that have a specific latitude and longitude associated with each update.  A third-party client on a iPhone can easily do this by querying the iPhone’s GPS system.

Why would this be helpful?

Here’s a quick example:

During the Atlanta gas crisis, users on twitter started using the #atlgas tag to identify gas stations that weren’t empty.  The logical next step would have been to create a map of these tweets.  But, since the locations were being written in the tweet, it was a serious challenge to accurately parse the messages and auto-create a map.  If each of the tweets accepted lat/longs, it would have become a trivial exercise to produce an extremely helpful map.

Obviously Twitter has a lot on its plate but I continue to believe that it needs to do a better job of making its existing data set more useful to non-Twitter users.  Adding more meta-information to each tweet would certainly help those third-party developers build more interesting applications.

Why Twitter Won’t Go Mainstream But Will Still Succeed

October 9, 2008 | Comments Off on Why Twitter Won’t Go Mainstream But Will Still Succeed | technology

I’m sorry, but my “normal” friends don’t get Twitter. They got Facebook, they got YouTube.  But, when I show them Twitter, they have no idea why anyone would use it.  The few of them who are social enough to broadcast short messages like to do so privately and to their friends (i.e., Facebook status updates).  So, is Twitter done?  Not at all— it’s just getting started.  Twitter doesn’t need to worry about getting everyone to start broadcasting messages, they need to focus on making their amazing data useful to everyone else. That’s what YouTube did.  YouTube succeeded not because it got everyone to contribute videos but because it took the videos of the few and made it useful to everyone else.

A few weeks back I created a graph, based on Twitter data, that helped people determine what time they should try to eat lunch at Shake Shack.  Really simple exercise but I got emails out of nowhere from friends that have never even heard of Twitter saying they were forwarded the chart and how useful it was.

What I did learn from my experience is that making twitter data useful is difficult. It’s not structured or organized and it’s hard to imagine how Twitter will ever get it’s users to structure the data themselves (hash signs will only go so far).  In other words, Twitter needs to do it themselves or someone needs to do it for them.  Twitter is headed in the right direction, the purchase of Summize to provide a Twitter search engine was fantastic and the new Election ‘08 page is interesting but not incredibly useful.  There’s way more to be done.

In terms of practical advice for the Twitter folks, I recommend they talk to the really smart guys over at Pluribo. They are using cutting-edge artificial intelligence to summarize Amazon product reviews.  Perhaps Twitter could encourage them to focus their time on the tweets fire-hose.  Imagine typing a phrase (like the recent debate or a movie) on Twitter’s search engine and getting a summarized view of thousands of people’s thoughts — pretty interesting.

Your Twitter Followers Aren’t Real

September 18, 2008 | Comments Off on Your Twitter Followers Aren’t Real | data analysis, technology

Based on a random sampling analysis of twitter accounts I conducted, 6 out of 10 twitter followers aren’t actually following you.  That would imply that Barack Obama, who has the most twitter followers at 80K, really only has 30K “real” followers.

I decided to take a closer look at the top three twitter tech-heavyweight (figuratively speaking) bloggers based on Twitterholic’s top 100: Mahalo’s Jason Calacanis (#7, 34K followers), Scobleizer’s Robert Scoble (#8, 34K followers), and TechCrunch’s Michael Arrington (#13 at 25K followers).  Even though Calacanis has a slight edge on Scoble, looking at their “real” followers was a completely different story.  Robert Scoble has significantly more “real” twitter followers (13.6K) than Arrington (8.6K) and Calacanis (7.5K).  On average, they were reaching 68% less twitter accounts than their follower counts indicated.  This isn’t a comment about them, they are fantastic.  It’s a comment about how twitter follower numbers are misleading.

Twitter users are pretty proud of their follower counts and they put it on their blogs next to their RSS reader counts.  I’m pretty proud of my twitter account and I only have 57 followers.  Twitterholic even puts up a leader-board of the top 100.  But, the not-surprising truth is that like RSS reader counts, not that many people are actually reading what you are tweeting.

As Twitter continues its impressive expansion and twitter accounts start to become businesses, it will be important to have a more accurate view of the reach of specific twitter accounts.

Several services are making progress on this front (Twitter Grader, Twitterholic) but there’s a lot more to do.

Note:  For the purposes of this sampling, I defined a “real follower” as someone who follows less than 300 twitter accounts and is active as measured by having a status update submitted in the last 3 days.  It’s definitely not a perfect definition but I hope it was good enough for the purposes of this demonstration.