The Virality Experiment — My Word Cloud

Originally posted on tokudu.com on 07/03/2011


Virality works. It’s a fact. It’s the core business component of many modern startups. It’s their key to retention, growth and, thus, success. 15 years ago, it took Hotmail almost a year to grow to 3 million users. In the past year, startups like Kik and LiveProfile did it in weeks and days. How? The answer is simple -viriality.

For a while, I’ve thought of virality as a mythical term that douchey business guys threw around to get investors empty their pockets for a new cool Web 2.0 startup. However, as I got more interested in entrepreneurship , I started to finally understand what it means for a product to have a viral component. Nevertheless, it was still something I have never experienced myself. Yes, I’ve seen products created by friends spread on Internet faster than high school rumors, yet I have never been the one to spark the fire.

It all changed over the last few weeks. About a month ago, I was in NYC on a work trip and I was hanging out with my good friend Ian who I haven’t seen in a while. In the past, I worked with Ian on TalkMesh, and he is definitely one of my favorite people to pair-program with. Since we haven’t coded together in quite some time, we decided to take advantage of my week in NYC and hack together on a project. Given the time constraints, Ian suggested to write a Facebook app with some sort of a viral component. Ian has previously successfully launched My Friend Map — a popular Facebook app for showing your friends’ location on a map. With my interest in virality and my past experience with Facebook APIs, his idea seemed quite interesting to me and well scoped for a single week of work.

After some brainstorming we decided to create an app that would generate tag clouds based on what has been posted on your Facebook wall. We looked at thecompetition, and realized that most of the similar existing apps really sucked at generating good-looking compact clouds. This is bad for virality.

A viral product typically has two notable characteristics:

  1. The “cool” factor or the “viral component”
  2. Simple sharing channel

While the competition also leveraged Facebook for an easy sharing model, they simply failed on the “cool” side of things. Let’s face it, people love to share stuf when it looks good. And good often means beautiful. It’s as simple as that. When people look at something that blows them away from esthetic point of view, all they want to do is share it with others to validate their own excitement. That’s why in order to make a viral tag cloud app, the results needs to look stunning.

So, what does a stunning tag cloud look like?

After some more Googling around, we realized that the most beautiful clouds on the web are generated by Wordle. This tool is quite amazing. By taking advantage of a great set of fonts, well thought-through color combinations and a smart generation algorithm, it manages to output some beautiful results. Even when the tags themselves were meaningless to me, the generated clouds looked so good that I wanted to share them. That was exactly what we needed it. Add the elegance of the Wordle-type clouds to the personalized set of tags extractable from your Wall posts, and you got the “cool” factor.

With that in mind, we started hacking away. Luckily, we were able to dig up a paper by the author of Wordle that helped us get started. A week later we had an app at our hands: I’m not gonna lie — I was pretty happy with the outcome. The results were quite stunning. Here is a quick sample:

The app’s functionality was very straight-forward:

  1. Go to apps.facebook.com/my_word_cloud/
  2. Customize your cloud by selecting a color scheme, a font and entering a caption for the image
  3. Generate the cloud and post it to your wall.

However, these 3 simple steps contained an important viral loop. The app added a caption to the image that included the URL to the app itself. Then, it posted the images to your wall for all your friends to see. Since the images looked really attractive, your friends would follow the link in captions to generate their own clouds. And so on..

We hoped that given the viral loop, the app would “explode” and spread like wild fire. And so it did.

Even though, we only seeded the app by asking a few close friends to try it out, two weeks later the app grew to over 200 hundred thousands users! The real growth happened over just a few days. Have a look at the graph below:

The most interested part of the growth was that the app somehow randomly exploded in Nederlands and Brazil. People there loved it:

What happened next? The app got shut down by Facebook. Apparently, we were violating some terms. They offered us an opportunity to fix things and relaunch the app. However, in my mind the project has accomplished its primary goal: it has proved me that virality is one of the primary drivers of product growth and adoption. If you are designing a product, you need to focus on the user. However, it’s often not enough. In order to succeed and succeed fast, you need a good distribution channel. Viral growth is often the solution.


Originally published at tokudu.com.