1. Basic quality. Your game needs to look better than games that can be downloaded for free. If it just looks the same, why should people pay for what they can get for nothing? The same goes for bugs. And length. If you are asking for money, people will compare your game with other games that they could buy. A short game, an ugly game, or a buggy game will fail. Simple as that.
2. What's so special about your game? There are thousands of games out there, and a lot of them are free. Why should someone pay for your game? If you can't answer this simply and convincingly, your game will not sell.
3. Distribution: to make thousands of sales you need to reach millions of people. And those people need to trust you. Just having a web site is not enough. The best way to reach people is to find an established games publisher to distribute your game. If you can't find a publisher to take you on, ask yourself whether your game is really so special. Most independent game makers admit that their first finished game is not as good as they first thought. But the next one is usually better.
   

How many copies could you sell?

You can't expect to sell as many as the latest EA games of Ubisoft blockbuster, but you have one big advantage: you don't have their costs. So you can make a profit on much lower sales. Even a small publisher needs to sell around ten thousand copies to break even. As an independent developer, you can make a comfortable living on half of that.

50,000 - 100,000 sales worldwide is a big hit for a small publisher in any genre. 150,000 gets you in that month's top ten games chart. Only about 30 PC games of any kind ever sold a million. (If we include games on consoles then the number is higher.) Myst is the best selling adventure game ever, and sold between 4 and 9 million copies (I have seen different figures from different sources - possible some numbers included sequels or console versions).

The most recent complete figures I have are from 2003, from PC Data, courtesy of justadventure.com. The appendix has the full numbers, and the previous two years for comparison. The games that sold the lowest numbers were more successful in previous years. The key points are these:

• Every single game sold at retail sold four thousand copies or more worldwide All of them. Yes, even the bad ones. Not all of them sold that many in one year, but they all sold that number in two years or so.
• About forty percent of these games sold more than ten thousand copies. One in three of these was a new title, not just a "safe bet" sequel or TV spin-off. So there is hope for brand new titles. The biggest selling title sold over 100,000 copies.
  
• Remember the economics of the video game market? Four thousand copies over two years is enough to live on, assuming that your costs are minimal. If you get your game into the shops, you are pretty much guaranteed that level of sales or higher. But has any amateur games maker done this? Let's look at some case studies.
  

Gilbert Goodmate (2000)

Daniel Nilsson and friends made Gilbert Goodmate, a game in the classic Monkey Island style, and released it in the year 2000 at a selling price of £29.99. 1,400 people downloaded the demo before the release. According to this publication, the PDA edition of the game has been downloaded 6,939 times by April 2004, at a price of £ $19.95 (about £12) a time. In other words, the PDA version alone made around £80,000 gross. I don't know the PC sales, but based on interviews and other evidence I would guess it sold over ten thousand copies.

Gilbert Goodmate had very good sales. How did they do it? First, they spent a long time getting it right, then testing and improving. They produced a quality product. And they got an agent, Clearwater Interactive (in the UK) who found publishers in different countries. They took the time to get it right.

So did they make a ton of money? Er, well, actually no. According to an interview at GameSpy they started with an investment of $ 380.000 to pay for several people working for several years. This is much less than most professional games, but a lot more than any amateur game. It would have needed around 40,000 sales to break even. By the time the game was ready, the investors got scared. According to messages here, Nilsson and co. had to form a company in order to get the investment. When the game was ready, the nervous investors changed the name of the company (from Prelusion to Global Minds), and declared it bankrupt. This allowed the investors to keep every penny of any sales, but it meant Nilsson and co never saw a penny more.

The moral of the story is that the initial investment was far too big to guarantee profits. If they had made the game on a much lower budget it could have been a huge financial success.

Dark Fall by XXv Productions (2002)

Dark Fall was made by Jonathan Boakes, who developed it at home, and was adopted by the Adventure Company. Hardcore gamers loved it (possibly influenced by the knowledge that it was a one-man production) and gave it high scores. Other games reviewers called it "sub par" and "repetitive" and gave it low scores. According to Metacritic, the average score was 68%. Dark Fall sold around 40,000 copies worldwide (extrapolated from the 18,000 sold copies in the sample used in the appendix of this article).