Do you have rules in place for doing something manually vs. creating an automated system?

I've heard from many successful entrepreneurs that only your job as a startup founder is to constantly be creating systems; However, There's a huge trade-off though between trying new experiments and being lean vs. just getting the system in place to not worry about it anymore and moving on.

Do you have a rule in place at your company for when to create a system or when you should just do it manually? What do I mean by systems?

• Training, hiring and managing a new employee to do a job
• Building computer systems to automatically do jobs

My startup is a small 3 person team so that is what I'm basing this question on. I'm curious as to everyone's thoughts no matter what the size of your company though. Thanks!

The Book "Rework" by Jason Fried and DHH of 37signals has a chapter entitled "Hire when it hurts". In other words, don't hire until the pain of not hiring is too great. I think the same principle applies to automating. Don't automate until it hurts. In either case, it may hurt immediately. Ok, then automate/hire immediately. Don't get enamored with creating systems for the sake of creating systems. Create systems only to solve actual, right-now problems and you will be fine.

My startup of two does not have any rules in place for this. When I make a decision on automating, or even training externals so I can buy something as a service, I consider it a decision on whether or not to invest in infrastructure. It is basically the same decision as whether or not to build a road from one place to another. Whether to do it is both a decision which considers the ROI, but also a strategic decision on whether you want to conduct business in the area, the new infrastructure enables you to.

On a different note, I have almost completely automated my invoicing procedure, even though it only takes up something like 0,01% of all my time, but it is just so important to get it done, and as early as possible, so I believe it was worth automating.

1. Do it manually, until you understand it enough to know what can be done automatically.
2. Automate things you really hate. I could play basketball for hours but dread 5 minutes behind a lawnmower.
3. Automate things you are not good at. This could overlap with #2, but not always. Be honest about your abilities and remember: having things donw the 'way' you want doesn't mean they're being done well. Perfectionism is a flaw.
4. Have a simple method of checking on the automatic processes to make sure they are being completed.

Document the way things are done and not they way you think they should be done. Schedule a time to review documentation. Many are afraid that documentaion of procedures can lead to inflexibility. This is obvious when you are doing something new for the first time; nobody wants to deal with a beaurocracy, but you come back to an existing system and need to make a change, lack of documentation can bring about, "We've always done it that way, so don't change it." You want to know why things are being done to help you understand the consequences of making changes.

Automation is a process that needs to be continuoulys managed or it will easily get out of control. And isn't that really the point?