Obscurity is Your Friend

Last month during the Facebook privacy backlash, Jason Fried from 37Signals wrote a great post about Diaspora, the “open source” Facebook that raised $200,000 on Kickstarter before even beginning their project.  They received publicity from mainstream media outlets like the NY Times and the fund raising took off. His argument, one that I agree with, is that Diaspora is taking the wrong approach:

Diaspora has all the wrong things at the wrong time. Competition that kills isn’t pre-announced — it catches an unsuspecting incumbent by surprise.

In particular, I found one of his points to hit home:

The spotlight is on too early
You want attention after you’re good, not before. Obscurity is your friend when you’re just starting — especially when you don’t even have a product yet. You don’t need the pressure of outside opinion or the press breathing down your neck before you have anything to show. Millions of eyes — including your competition — watching you every step of the way doesn’t help. All this attention is a distraction. Ship, then seek the spotlight.

I’m going to go out on a limb and say that every single piece of software – web or otherwise – that has ever been released in the history of computers has been buggy at first. We’ve had a pretty disastrous start with LockerPulse from a product standpoint. We’ve had every potential hardware and software problem that I could have imagined. The site has been unstable as hell.

Some people might look at this as a bad thing, however, because we did a small “beta” launch, we were able to get real users on the site, get real feedback, observe what worked and what didn’t, and then begin to fix things. You’d think that because we’ve launched a lot of websites over the years that we wouldn’t need time for this stuff and that we could just go full blast with the site from day 1. It just doesn’t work that way, especially with software almost 100% built in-house that has never been used in a live environment before. It has been a little worse than I had expected, but then again I wrote this just before launch:

We’re going to do a “public beta that we’re not calling a beta” for a few months and then go from there. What does that mean? We’re going to be giving out free lifetime premium accounts to a bunch of people we know (friends, their friends, family, people we know in the web business and sports business), we’re going to be testing a micro social media campaign to gauge it’s effectiveness (we think Twitter is maybe our best marketing opportunity), we’re going to be soliciting feedback and collecting data, and then we’re going to go from there. We have a laundry list of things that we want to do, but none of that matters if the first 50 people that try it think it sucks. This is an app on a level that I haven’t built before, so it’s important to make sure we didn’t royally screw anything up before going too crazy. Right now it looks like we’ll be adding 50k – 75k stories per month to our database. There’s a lot that goes into making that work efficiently.

Now, some 40 days later, we’ve worked out some serious coding issues, made some changes to appease some of the major content providers, completed a UI redesign, and moved to a much better server*. We’re getting close to the point where we have the kinks worked out and we’ll be confident spending some time and money to market the site. Imagine spending thousands of dollars for some big huge launch and then realizing that your server couldn’t handle the site or that half of your users couldn’t use your interface?

“Obscurity is your friend when you’re just starting.” I love that quote from Jason. This past month has been tough on me, but it would have been much worse if all of the problems were exacerbated by gobs of traffic. If the site is slow/down, I might get an email or two. If I got a few hundred it would be a different story. Then you’re playing disaster control with your customers AND trying to fix a bug.

Big huge launches are sexy, but they aren’t very practical, especially for a small bootstrapped software project. Slow and steady wins the race.

*Originally I had picked out a VPS to start development. I thought we could launch and be OK in that setup for a little while, but it became evident that we needed more power and even more flexibility should we grow the site quickly (no one wants to fear being too good at marketing, it’s hard enough as it is). Our hosting provider, LiquidWeb, who we’ve been with for years and I’m very happy with (no downtime…ever), worked with us to transition the site to their new cloud product, Storm. This gives us the ability to add RAM or a CPU or even a load balancer with a click of a button, something I think is necessary for a site like this. I had originally considered Rackspace Cloud and Amazon EC2, but Rackspace had too many downtime horror stories (see TechCrunch), and Amazon didn’t have a level of support I was comfortable with. Cloud servers are the way to go for web apps now a days, so I’m happy that LW is finally offering an option that works for us.

8 comments on Obscurity is Your Friend

  1. Amber says:

    Did you actually experience issues with the VPS or was this a preventative measure? How has the all-in pricing for your VPS compared with your cloud usage?

    • Adam McFarland says:

      Good questions Amber. We did experience overloading at times, but it was mostly because of our back-end stuff. “Polling” RSS feeds (checking them every 15 minutes for new stories) is very resource intensive. The site itself is optimized pretty well with a lot of caching going on so there’s not a ton of resources being used there. However, the polling was slowing down some of the premium features, and on occasion crashing the site if there was a lot of users plus a lot of search spiders. Price wise, I love how Storm is set up. You can do more of a pay-per-use thing, but you get a lower price if you “block out” a set amount of bandwidth for yourself. Our cost went from ~$100/mo to ~$150/mo to go from a single-core 1GB machine to a dual-core 3.5GB one, with better backups and more scalability. As soon as we moved you could tell the difference. No queries getting backlogged or anything. Even a backup wold slow things down on the VPS, but those run instantly and very smooth now. And, if we need 8GB of RAM or a quad-core, it’s just a click of a button 🙂

  2. Tim says:

    Ironic, late last week we converted our site from a VPS server to a Storm Cloud server from liquid web! I didn’t realize that is the direction you went as well. The packages they offer are very flexible and so far the support has been as good as anyone else we’ve dealt with. Our programmers STRONGLY suggested going with a more “serious” server or we would run into problems in a few months, it seemed like the obvious thing to do. Our old developer has had so many bad experiences with Rackspace they told us they would turn down our project if we insisted on using them and if we went to them they would no longer offer support. Factor in the INSANE price and there are way too many other good competitors to take the risk, even with huge cash reserves it just doesn’t make sense.

    • Adam McFarland says:

      I did not know that Tim. That’s awesome! I think you’ll be very happy with LiquidWeb. I’ve had 1 bad ticket with them in 2 years. It was recently with the VPS, but the end result was a move to Storm, so I can’t be too upset. The rest have been 10/10 help, even on issues that were outside of their boundaries (say, adding indexes to our database when they noticed it would improve efficiency). Also if you call you get a sys admin directly, which can be helpful in an “emergency”.

  3. Rob says:

    Until you’ve got something worth shouting about… don’t shout.

  4. […] I mentioned in June, obscurity is your friend. It took over a month to get right, but by seeing the true technical demands of the live site we […]

  5. […] RSS for new stories, to delivering them to our users, has needed a lot of work. We also had plenty of server issues early on. Not that things are perfect now (far from it), but we have enough information to say […]

Comments are closed for this post.