2

Politician Market – Post Launch Writeup on Conversions, AppEngine, and Launching

On Tuesday, I released Politician Market. I brought it online, posted to Hacker News, and went to take a shower. 15 minutes later, I’m wearing a towel when I notice that Politician Market is #1 on HN and the server is melting. 12 requests per second. Gosh. I spent the next hour coding in underpants only. My roommate walked in with family members in the middle. He just pointed, said something with the word “roommate”, and walked out.

I wanted Politician Market to be entirely static site, so I used JotForm for signups. There were so many sign-ups that I crashed JotForm’s servers after only 10 minutes and had to move the sign-up form to Google Docs.

Signups and Conversions

Conversion tip: If you want to optimize sign-ups, don’t use  an external service. Sign-ups stopped almost entirely after moving to Google Docs. (I received only 40 signups after the move.) If you want to gain feedback, do place a feedback form on the homepage and you’ll get plenty of feedback.

12,000 uniques and 4 hours later, Politician Market mysteriously disappeared from the homepage and traffic started to subside. Some time before then, I had to disable the contact form entirely because I was about to overrun my JotForm quota. Despite disabling the form, people kept submitting it, often blank. (A browser bug?) At that point, I changed the form’s action to “#” and added an overlay (see below), but people kept submitting the damn form. My brother told me that he wrote me a long message in the contact form, because he was convinced that it really did get sent after all. What must you do for people to believe you?

Another conversion tip: Adding social sharing buttons seems to matter. There were 12,000 hits while we were on HN for four hours. After pg killed the submission, the long tail brought (and is still bringing) another 2k. If you add sharing buttons, make sure to include Google Plus. The break down was 259 shares on Twitter, 40 on Google Plus, and a bunch on Facebook. (For an unknown reason, Facebook wont show me the stats anymore.)

I didn’t have time to A/B test, but conversion rates seemed to rocket when I placed annoying sharing links so far up that they pushed content below the fold. I hated that, so I soon reverted the changes. (Startup idea: I wish I could run automatic A/B tests by including a javascript file that would hash the page, check with a 3rd party server to see if it changed, and provide conversion data when I’m cowboy coding. Create that and I will pay a one time $10 fee for the service.)

For absolutely no reason, I asked for phone numbers in the signup form. About 30% of all sign-ups included them.

Avoid AppEngine Like the Plague

I originally wanted to host on S3, but it wouldn’t accept my credit card so I used AppEngine and static-app-engine-hoster. Don’t do that. There is no caching and you lose all the benefits of a static site. I wasn’t expecting so many hits, so I didn’t think this through.

More on AppEngine: Deployment is easy, but payment sucks. AppEngine makes you pay per-week, so if you want to use $10 a day, you have to deposite at least $70. (Slimeballs.) AppEngine has absolutely no support for paying customers. I sent two emails about urgent issues and never heard back. Avoid AppEngine like the plague.

Publicizing

I tried publicizing Politician Market on Reddit, Slashdot, and Digg. It almost took off on Reddit, but Slashdot ignored it despite votes on the Firehose submission and Digg was a laughable waste of time. (What they say about Digg is correct: If you aren’t part of the oligarchy you might as well submit the link to /dev/null.) I emailed a few tech and political bloggers. Most of them ignored me, but some had incredibly kind words and published a link. Do have the chutzpa to email bloggers.

This is the second launch I have done. Third time ice cream. Does anyone know where that weird Israeli saying comes from?

5

Why Yahoo is Irrelevant

Earlier, I ranted about the poor quality of answers on Yahoo Answers. In 2006, information on Yahoo Answers was of higher quality, and the site was hailed by TechCrunch as Yahoo’s long term strategy for overtaking Google:

Yahoo earlier this year oddly admitted that it’s not their goal to be No. 1 in Internet search. “We would be very happy to maintain our market share,” CFO Susan Decker told Bloomberg… I think that they decided to lose the algorithmic keyword search battle and focus their resources elsewhere in order to win the search “market share” war…

Yahoo! is sensibly looking change the battle ground by providing a different type of search, one not based on complex mathematical algorithms but one that could answer questions like “how will mankind save the planet?” Putting this question into Google resulted in no search results being returned and yet you could pose the very same question on Yahoo! Answers and would probably get a few hundred answers if you were not famous and several thousand if you were.

What went wrong?

If you read Forbes, you may remember a 2010 article, titled Stop Focusing on Your Core Business, which suggested that Yahoo had failed because they couldn’t move fast or enter new markets. I can’t understand how anyone can think that, considering that even this run-on sentence only covers 25% of the 72 businesses that Yahoo has ever been involved in. Rather, Yahoo’s mistake was utter incompetency and a chronic lack of focus. They focused on width over depth, and they lost everywhere as a result. (StackOverflow is the counter example: They focused on community and quality, and the resultant community is indisputably excellent.)

If you follow Hacker News, you surely know the most common advice given to startups: pivot. Pivoting is swell, but here is a more fundamental lemma that Yahoo would do well to study: Don’t make half-assed attempts.

A search for the word “pivot” on Hacker News returns 1,923 results, a search for “core business” only 594. That’s a shame. Google succeeded because they knew their core business was search, and they provided the best damn search results anywhere. If you’re creating a startup, or writing a piece of software, you should focus on competency in the narrow problem space you choose. There may eventually come a time to expand -  hell, if you’re there, you shouldn’t be reading this blog post – and note that most areas Google expanded to were related to selling ads (Gmail, Google Maps) or improving search results (Google Plus). They always focused on consumer problems and provided solutions that were better for consumers.

If you read until here, you should try Freeversation, my new take on online anonymity. With Freeversation, you can have private and anonymous conversations with friends and colleagues. You always know who you are speaking to, but not who said what. We have no plans to offer any unrelated services ;)

12

Project management software, wikis, and other technical infrastructure for small startups

Crucial infrastructure for small projects

My latest project is a web application, which is currently in closed beta. I will blog about that more in the coming weeks, but today I’d like to discuss my project’s infrastructure. Infrastructure is one of those topics that you usually just learn about from the companies that you work with. I hope that the information here will be useful to individual developers and small startups. It is difficult to know what has to be done when you’ve never been part of a real-world development team before.

My project’s team is made up of three people. For a project of that size, there are three resources that I consider to be absolutely crucial:

  1. A project wiki – For storing project information like budgets, marketing plans, etc. The wiki should replace most emails between team members. The point of a wiki is to establish one place to look for the latest information, so that you don’t need to search through old email threads to find something.
  2. A project management site – For creating roadmaps, planning new features, reporting bugs, and assigning work to each team member. If you’re going to ignore everything else I write in this post then just setup a project management site. You wont be able to go back and live without it.
  3. A code repository (aka version control system) – For keeping track of changes to your codebase. Also useful for storing all of your source code in one central and backed-up location. (Yes, that explanation is slightly simplified, but using a version control system should be a given.)

Depending on the nature of your project, you may also need a build-bot or continuous integration server. I don’t have any experience in that area. However, I hear good things about  Jenkins (formerly Hudson)Chef, and Integrity. (Edit: As Tollef Fog Heen pointed out, Chef is a configuration management tool and not a CI server.)

… Continue Reading