What does a software company cost to start?
How much money do you think it takes to start a software company? $10,000? $100,000? ONE MILLION DOLLARS?
A lot of startups talk about revenue, but not many talk about costs—operating or starting up.
Note: assume everything I ever write relating to software is about bootstrapping/not involving VCs. I’ve worked with startups who raised capital and it’s not better or worse than bootstrapping, but it is a different mindset when you’re spending someone else’s money. And spending an amount that you couldn’t provide yourself.
Obviously, costs can vary heavily. Some software takes a lot longer and is much more involved to create. Sometimes you need to contract out a lot of work because you simply don’t have a specific skillset to accomplish it quickly.
For ofCourseBooks, we decided we wanted to spend as little as possible to get it off the ground. Not only because we’re cheap, but also because we wanted to prove to ourselves and other folks thinking about creating software that it can be done, and on the cheap (< $1,000).
Another thing to note is that before we spent more than a few hundred dollars, we opened up “founding memberships” to early backers that could support our development efforts by buying lifetime access to ofCourseBooks for a lump sum of $200. We sold 63 spots, or $12,600 worth of revenue. This also helped validate the premise of the software, since people were eager to buy it before they had even laid eyes on it.
Let’s talk about skills for a second in this walk-through of expenses, because skills can offset costs. The more you can do on your own, as a founder or cofounder, the less you have to spend on contracting those skills out. Sometimes it makes sense to outsource skills if it’d take too much of your time to learn or even do certain tasks. But for larger ticket items, knowing how to do something is almost always cheaper than paying someone else to do it.
The main skills we used to create and now operate ofCourseBooks are:
- Application programming
- Server setup + management
- Design + branding
- Email automation
- Marketing + promotion
- Audio/video recording + editing
- Customer support
- Operations (this is often overlooked, but the ability to keep a team moving forward with lists and action items is huge)
It’s not just the skills that are important though, it’s also the overlap and how well they work together.
Here are the costs, to date, for creating ofCourseBooks:
Digital Ocean – $20/month
We have two droplets, one for the app and one for the front-end website, which handled ~ 10,000 visits in the first month.
Heroku – free
We use Heroku for a staging server for now, since it’s free and was the easiest place for Zack to code the app where Jason and I could see his progress and test.
Amazon Web Services – $.03
Funnily, we should be on the free AWS tier but there was an emoji-related fuck-up that required uploading the same thing a few times, pushing us to spend an ultra expensive $.03.
Hover – $41.12/year
We bought ofcoursebooks.com and ofcb.co (easier to share workbooks on a domain that’s 4 characters).
NameCheap – $18/year
2 SSL certificates. One for ofcoursebooks.com and one for ofcb.co.
Klim Type Foundry – $600
Almost ridiculously, our largest expense for creating this software was the fonts Calibre and Tiempos from KLIM. We wanted our app and site to be unique and stylish, and those two fonts pair perfectly together. I felt bad suggesting them, since they cost us $600 for the web licence, but they are perfect.
Stocksy – $52.50
Most stock photography is awful. That’s why I only buy photos from Stocksy – search for “Business” or “Marketing” on any other stock site and then again on their site. I guarantee one will show you the worst, cheesiest, awful images and the other, obviously, is Stocky’s site.
Venjuris – $375
We established an LLC in Florida with the help of Ruth Carter, Esq. at Venturis, because we wanted to make our partnership legit. And since I’m a dirty Canadian socialist, we were limited to where and how the company was setup. We even have a signed partnership agreement!
Ruby on Rails – free
The ofCourseBooks app is built on Rails, which is open-source.
WordPress – free
The website is built on WordPress, which is also open-source.
Helpscout – free
To manage support requests and emails, we use HelpScout’s free plan. It’s easy to use and really helpful to organize incoming emails that the three of us need to reply to. Plus their free plan allows for 3 users!
MailChimp – $9/month
We pay for MailChimp even though we have less than 2,000 subscribers (which their free plan would cover) because we need automation to segment out and send different emails to customers vs. subscribers. MailChimp handles all front-end/marketing emails. I teach people how to use MailChimp.
Mandrill – free
Mandrill handles all emails that come from the software (account creation emails mostly). We’re under 2,000 sends, so we don’t have to upgrade to their $20/month for 25,000 emails yet.
Google Analytics – free
We track every event in GA, from sales to button clicks, to referrers.
GoSquared – $9/month
We also use GoSquared for just the website because it gives us a quick snapshot of traffic, effectiveness of our campaigns and events like email subscriptions.
Papertrail – free
Papertrail logs error messages for us so we can pinpoint any problems and fix them quickly. So far we only need the free plan.
New Relic – free
This tool monitors how well our software is performing so we can diagnose and fix bottlenecks or processes that are eating up time + resources. We are currently on the free trial.
So, all in, it cost us $1,125 to go from an idea to a functioning software product (including our first month of operations).
Moving forward, we’re currently burning through about $45/month to keep our software operational. As we grow our customer base, that number will also scale—as will the work involved to support and add new features to ofCourseBooks. But still, at $45/month and $1,125 to start the company, means that even if we don’t make another dollar, we can stay operational for 20+ years.
Now I’m sure the nerdiest of folks reading this will be appalled that we chose a certain tool instead of another tool, or Heroku for staging instead of a synced droplet on Digital Ocean, but we made the choices we did to move quickly from idea to launch.
Things will definitely change and the tools that don’t serve us in the future will be replaced. But for now, as they say, this is how we do it.
The point of this article is twofold.
First, we like sharing the behind-the-scenes of building this software, since it was created totally in public on the Invisible Office Hours podcast (season 4).
Second, is that software doesn’t need to cost a ton of money to build, if (and it’s a big if) you scale down your first iteration to just the core features you need and follow that set of core features and ignore everything else until a further iteration.
Thanks to Luke Chesser and the Crew crew for posting this. It inspired this post and a re-think of simply sharing detailed revenue numbers but glossing over expenses.
By Paul Jarvis—who has more tattoos than you.