The value of a quiet 1.0 launch

As anyone involved in the creation of software can attest, shipping a product takes an unbelievable amount of hard work and coordination. “Flipping the switch” and releasing to the public is an exciting time, so the natural tendency is to try and hit the ground running with promotional efforts. Despite the popularity of this approach, it’s a mistake.

When we released the original version of Flixel back in March of 2012, we reached out to dozens of media outlets leading up to the launch. When FastCo Design said they were interested in doing an article about Flixel, we were ecstatic. The article was great, and we timed it perfectly with the release of the app. FastCo’s audience turned out to be a great fit, and within the first couple weeks we had a nice big flood of users coming in. 

How quickly a blessing becomes a curse.

A login bug prevented a lot of users from signing up. Others experienced slow loading times and crashes on creation. People have very little tolerance for roadblocks – as quickly as the masses of people came, they left. Sadface = us.

In an ideal world these things would have been caught in testing and our servers would have been ready to scale. Unfortunately the ideal world doesn’t exist – testing doesn’t catch everything, and the arrival of hundreds of thousands of people is tough to simulate. Like most new apps, we weren’t ready for prime time, and we got burned.

Lesson learned.

Last month we released an entirely new set of Flixel apps rebuilt from the ground up…and we didn’t say a word. Despite our confidence in the stability of the new system, we played it safe and just quietly released the app to make sure everything worked. As luck would have it, on the day we launched our DNS provider suffered a major DDOS attack. This combined with our migration from staging to production servers meant that a small percentage of new users ended up on our internal system. Certainly bad news, but problems are a lot easier to fix when they only affect a couple dozen people. We also found a facebook sharing bug, and a couple other issues that we’ve fixed in 1.0.1.

Despite all our careful planning this time around, if we’d done another big 1.0 launch, we would have been burned again. 

There’s a world of difference between practice and showtime. Stand-up comedians don’t try new material in the middle of their HBO special; they hone their act in small clubs until it’s ready. Software is no different – as the saying goes, you’ve got to nail it before you scale it.

With all that said, we’re happy to announce that Flixel 1.0.1 is live in the app store and looking solid. Thanks to our cautious launch strategy we’re ready to start our promotions and not losing sleep this time around.

Well, not as much sleep. 

12 Comments

Leave a Reply

Your email address will not be published. Required fields are marked *