Ask Slashdot: Best Practices For Starting and Running a Software Shop?

An anonymous reader writes: I'm a systems architect (and a former Unix sysadmin) with many years of experience on the infrastructure side of things. I have a masters in CS but not enough practical exposure to professional software development. I'd like to start my own software product line and I'd like to avoid outsourcing as much as I can. I'm seeking advice on what you think are the best practices for running a software shop and/or good blogs/books on the subject.

To be clear, I am not asking about what are the best programming practices or the merits of agile vs waterfall. Rather I am asking more about how to best run the shop as a whole. For example, how important is it to have coding standards and how much standardization is necessary for a small business? What are the pros and cons of allowing different tools and/or languages? What should the ratio of senior programmers to intermediate and junior programmers be and how should they work with each other so that nobody is bored and everyone learns something? Thanks for your help.

Comments

First and foremost

[vivaoporto]

Get a good accountant to keep the books in order. Get a good lawyer so you always have someone vetting your contracts and preventing or solving any litigation you may find yourself entangled.

Don't try to do all by yourself, delegate everything you are not a specialist so you can focus on your core aptitudes.

Re:First and foremost

[Nuitari The Wiz]

Also look at oursources payroll, time tracking (this is sometimes a must for R&D tax credit) and make sure you have some financing / funding lined up. You need to have a plan to cover the first 2 years of operations where revenue will be slim.
This will also allow you to avoid getting into the "anything for a buck" mentality.

Don't focus on development tools / standards. Let your programmers take care of that. You might want to look for a lead developer with experience managing junior / intermediate developers.

Re: First and foremost

[American AC in Paris]

Even before that: have a business plan. Do your best to determine what you want to create, how expensive it will be to make it, how many people you'll need to manage, how much you expect to charge for it, and how big your likely market is. If you discover that there's no way to make your endeavor even close to profitable, you can save yourself months of heartache and mountains of lost money. Always have a plan, even if you don't stick to it in the end.

Re: First and foremost

[drooling-dog]

It's always a good idea to have a rough map of where you think you're going, but be careful about getting too carried away with formal business plans. You'll meet lots of people educated in business who will tell you that you need to sweat blood over a comprehensive plan - to the neglect of everything else - and then tour the country with a finely polished road show pitching it to potential investors. They tell you this because it shines the spotlight on their own training and talents. In reality, successful software business development almost never works this way, unless you have a stellar track record with several big hits behind you already (in which case they're investing more in you than the specifics of your plan). As others here have pointed out, what matters most is your rapidly growing list of happy, paying customers. Don't let your focus get diverted too far from that.

Sorry, you'll have to outsource.

[BarbaraHudson]

I have a masters in CS but not enough practical exposure to professional software development. I'd like to start my own software product line and I'd like to avoid outsourcing as much as I can.

Since you're getting into something you by your own admission lack domain experience in, unless you've won the Powerball and have a lot more money than brains, anyone you interview will realize that you're going nowhere and hence even the short-term prospects are, at best, poor.

At least with outsourcing, you can BS them as much as they BS you so they won't walk out the door shaking their head.

Bonne chance, 'cuz you're gonna need it.

Know what you're going to do

[wvmarle]

The summary is a bit short on detail, but one thing is lacking: a business plan. I've never run a software business, but am running my own business now so have a bit of experience in setting it up. What do you want to program? Who are your customers? Where's the demand?

You're already talking about hiring multiple people - this means you must have a decent outline of a piece of software to develop, and it's going to be a quite big project. Do you have customers for that already? Without customers, you're going to run dry very very soon, and you won't be able to get any funding. No customers, no future for whatever you want to do. Just saying "let's set up a software shop" is a one-way street to bankruptcy. You need to have potential customers before you start producing anything, really. You need to know the demand is there. You need to have your income sources. You'll have to find customers who need a product, and who believe you can deliver what they need at good price and quality.

Hiring people is very expensive for a shop without income. I've always started up on my own, do everything in house until you have too much to do that you have to start getting other people involved. In the meantime this also means that revenue is there.

Getting started is hard: no-one knows you, and hiring you (the new kid on the block) for some big, expensive software project (the kind a single person can't handle) won't happen. They'll go to the ones they know that can handle it. You'll have to start small, slowly get your way into the market, get your name out, get your product out, let the people know you're there and you're good. Then you may get bigger projects, then you may start hiring people and setting up an office that's not part of your living room.

Good luck with it all!

Did that, a couple times. Jump 1 employee to four

[raymorris]

That's what I've always done, grown each business slowly, organically. I've since learned that there are two types of companies that work well - tiny ones that basically provide the owner with a job, and larger ones run by a management team.

    What I did for far too long was deal with payroll, unemployment taxes, health insurance, sick leave, etc for two employees. That was a mistake. I should have chosen to either stick with just me and a part time helper, or make the jump to six or eight employees. That jump requires a leap of faith, some investment and a marketing campaign. Not making that leap meant that the business was dependent on one or two long -term employees who occasionally get sick, leave the company, etc.

Be tiny for a while until you figure out what you're doing. That may mean doing your business and a day job for a little while until the business provides you with a full-time income. Once it pays you $60,000 / year, then decide to either stay at that level or increase revenue by 500% quickly. Especially after the changes in the last six years, being an employer takes a lot of time and effort. Make it worthwhile. Do a POC by working it by yourself first, though.

Class projects vs. professional projects

[Anonymous Brave Guy]

The pay cheque isn't the important thing. Experience working in a professional environment is. The difference between how you work on a class project and how you work in a professional environment is vast.

For example, class projects are typically:

- very small

- implemented by a single person or at most a very small team that does not change over the lifetime of the project

- finished within a short period of time

- built with unchanging requirements determined by a single authority and entirely known from the start

- implemented with little need or regard for ongoing maintenance.

Exactly none of those things will be true of a typical industrial software development project. The need to take these kinds of factors into consideration completely changes how you design your software, what tools you use, what processes you follow...