Agreed. I had a project a couple years ago, just like this. Everything started off fine -- solid contract, good communication, etc. Payment was solid for the first few months, then it stopped. I was accused of putting together templates with no functionality behind them, and they demanded their money back. When I informed them where they could put their demands, so to speak, they threatened litigation. I had done the work and knew the contract was solid, but I couldn't have afforded a lawyer. Thankfully, they soon disappeared... But not before stealing their customers' funds. Ugh.

In retrospect, there were a few warning signs fairly early on, but it's hard to see that when you need the money to eat.

We invoice weekly. Depending on our terms with the client, if they aren't paying on time we stop work. This results in at-most 2 weeks of unpaid work, and typically they have a lot invested in seeing it through to completion because already have (largeish-n)% of their money. We also send emails EVERY TIME a work unit is created in xrono (our open source app-to-run-a-software-consultancy) showing both the current work and their current uninvoiced total.

Our clients never, ever, ever tell us that they were unaware of how much money they'd agreed for us to spend without us laughing a lot. It typically ends well.

"The real problems are the charming, smooth talking, dream clients offering cool projects, top dollar, and paying on time...until they don't and decide to start making your life hell because you're in so deep already. They're no dummies."

I can completely relate. If only there were a place to keep track of clients and their payment history. Glassdoor meets credit reporting. Clients with no history get charged more and require a higher deposit.

I agree with what you said, even though I luckily still hadn't any client like that.

I think, though, that a good method is to ask for an advance and maybe payments as the work progresses. I guess bad clients never pay advances.

Fortunately, 90% of bad clients are the former.

Doesn't match my experience: I have several small clients who pay within a couple of hours of receiving a bill. However they are profitable small clients (and very nice people).

I stopped making websites for friends or 2-person companies for this reason. Besides the pain on their faces when I estimate them how much its going to cost, these are the same people who get an idea in the middle of the project and want to add a dancing and singing cat to the home page.

- I agree, I've had bad experience with both. Be wary of individual/companies who work for clients and outsource their technology to you. Often the specifications will be very vague, and even if you manage to communicate a better level of understanding what THEIR CLIENT actually wants (which takes time and effort..) - you might end up not getting paid, because their client doesn't accept the solution, and doesn't want to pay the guy who hired you.

- Always have signed agreements, ALWAYS, can't stress that enough. Unless you have very good reasons of trust. Chat logs, e-mail conversations, all cool, all legal, but you will always get a yes-no story. Signed agreements are often very clear cut.

= Never agree on doing a revenue share with individuals or smaller companies. They probably won't have a clue what they're doing.

- If you don't believe in the product when you get your first impression on it, you never will. Don't let 'em fool you and convince you otherwise.

This is a good rule of thumb, but the best client I ever had was in the general area of legal services. No more than 6 employees, or at least that's how many email accounts I set up, but paid well, without complaint, and always immediately.

I think it comes down to, looking out for clients that focus on cost instead of value.

I would love to stop doing work for small companies or individuals. But how do I get bigger clients? I really don't know how.

Only half? For Craigslist this sounds on the low side.

Pattern matching is a general term for what you do every time you see a letter and recognize the pattern of light, or hear a word and match it to a concept. Don't give it a bad name just because human pattern-matching is flawed and unreliable in the (relatively rare) edge cases.

"Pattern matching" is just another phrase for applied Bayesian probability.

Maybe you can drop the "for smart people" -- the same kind of magazines that blather about horoscopes tend to run lots of stories along the lines of "How to recognize a narcissist".