Ask HN: Is the concept of a one-person software shop still viable?

Too many pay apps and even games with DLC have gone poof in recent years because of the latter situation or because the cost of maintaining the DLC no longer justified itself.

Just build something that you know people need (meaning: there are other products in that space) that has a small enough scope so a single person can build it in a month or two. Then charge money for it.

The software market is ludicrously large, and because you don't have marginal costs in a software business you can afford to make a lot of mistakes on the business side of things and still make a profit.

You don't need a great product (plenty of lousy products sell well) and you don't need great business sense (many CEOs don't know what they're doing) and you don't need great marketing (same). For your business to work you have to clear the minimum bar in all three areas, but luckily the bar is set pretty low!

It's not easy though. Assuming you have product/market fit (i.e. something that sells):

Reliability is paramount. It is everything to you. If your system breaks, it will stop new development, marketing, support responses, sales calls. Because there is only one of you, you can't afford to spend time fighting fires and answering pages. It does not matter if your software is slow or fast or if it is pretty or ugly or if your userbase is growing or not if you are down.

Reliability is paramount. Invest your time into building systems that do not fail when one component or one machine breaks. Where you can, you should leverage primitives and services from cloud providers that provide the kind of failure guarantees you need. Then assume that your cloud providers will eventually fail on you, so design with that in mind. Take as few dependencies on them as you can handle or make sure you can failover between them.

Reliability is paramount. Every change you deploy could break your systems and cause downtime. You need good testing and monitoring. Run continuous end-to-end testing of all customer-facing functionality that pages you when it fails.

Reliability is a feature if your software is critical to your customer's business. Some will notice when your competitors are down and you aren't. For those that don't notice, educate them. Explain to your customers the investments you've made to keep your service up and running. You can sell it as a differentiator.

After that, support is the next time-sink you need to eliminate. Treat every support request as a bug that can be fixed so it doesn't happen again. Make it extraordinarily easy to contact you and then try to optimize so no one ever contacts you. Invest in your UX. Your UX should try to illuminate the inner workings of your software. Many support requests are simply failures of a customer to understand what your software is doing and why. Customers can't debug black boxes, but they are smart and motivated and if you invest in their understanding, many will solve their own problems before contacting you.

Design your error messages. Your error messages are a more important piece of design than any other messaging from your product. Finally, if you can't solve the support bug with UX, invest in your documentation. Documentation is your last resort because most users will not read it or they will only read portions of it. By the time they get to the docs they're already frustrated, so it needs to be fantastic.

good luck!

I have a full time job but one client on the side. I built a system for them that I net $1,250 per month on and do 2 hours of work a month at most. Rackspace takes care of almost all host related items for me.

I have a partner who also makes $1,250 from the same contract. In hind sight I could have easily done the whole project without him and be making $2,500 per month.

Been going for almost 3 years now, probably has a shelf life of another 7 years.

If you can get a few customers like this, you are set. The challenge is of course to get more customers, let me know when you solve that issue, I'm still working on it :)

edit: changed gross to net