I remember talking to friends about how many copies they thought some game would sell. Often times, such discussions would include "Oh, it's a great game, it'll sell millions of copies."
You need a very good game to get a million units. But it doesn't mean your game is necessarily better than some game that sold half that or a tenth of that. It just means you had better distribution.
One of my favorite games, Dominions 2, probably didn't sell very high volumes. Does that mean it wasn't a great game? No. It just means it wasn't massively distributed. By contrast, when I walk into Best Buy and find literally 40 copies of Oblivion on the shelf on the first day, you know it's going to do some serious sales. Combine with the fact it's a really good game and you have a mega hit.
Quality of game is a major factor in determining how close you'll reach your sales potential. But it is distribution that sets that potential.
For example, below is a sample of the top 10 best selling PC games from February (before GalCiv II shipped):
- The Sims 2: Open for Business
- World of Warcraft
- Star Wars: Empire at War
- The Sims 2
- D&D Online
- Lord of the Rings
- Age of Empire III
- Civilization IV
- Zoo Tycoon 2
- Battlefield 2
Many of these games are fantastic. However, their quality does not necessarily translate to sales. To be in that top 10, you have to be in a LOT of channels AND have a good game AND have a game that appeals to a very wide audience.
The first Galactic Civilizations (2003) initially shipped into EB and Gamestop. That was it. A couple months later it got into CompUSA and BestBuy and a few other places. Obviously, that limited the title's potential in sales. We were happy with the sales, however. Our expectation wasn't to sell a million copies. We only expected to sell 30,000 copies with the hope it might do 50,000. It ended up doing around 150,000 total worldwide.
For Galactic Civilizations II, we self-published and put more effort into getting into more channels on the first day. This had terrific results. Our goal with the sequel was to do around 200,000 total sales with a higher % of those sales being full-price sales (a lot of GalCiv I's 150k sales were deeply discounted sales due to our publisher deciding to discount quickly and heavily in exchange for up front advances). As I write this, GalCiv II has already exceeded the # of units GalCiv I shipped in North America.
A lot of that difference was distribution. But not all of it. We've already sold more copies of GalCiv II direct than we did of GalCiv I total. We're still trying to understand why that is. Is GalCiv II that much better than GalCiv I? Are more people just willing to buy on-line now than they were in 2006? Was our marketing just that much better? It's some combination of factors but how much of each factor we don't know.
At retail, distribution sets your potential. People are buying a box. The more places that have your box, the more sales you can potentially make. For instance, Galactic Civilizations II has been in the top 5 (or even #1 in some cases) in the stores it's in -- Best Buy, CompUSA, EB, Walmart, GameStop, Amazon.com at various times. And those are only in the stores that we can access rankings. If we hadn't sold out, our stats would probably be even more impressive (first time indie publisher not easy to get retail space).
Needless to say, we've been really thrilled with sales. If a little company like Stardock can be #1 at Walmart or BestBuy or EB or whatever, then anyone can do it -- if you just make a game people want. It really demonstrated the power of word of mouth.
But we'll never sell a million units or anything like that. To do that, you have to be in a lot more channels. We're in Walmart. That's good. But we're not in CostCo, or Sams Club, or Target, or Office Max, or tons of other stores. I walk into my local comic book store to pick up the latest Ultimate Spider-Man and there's World of Warcraft for sale on the shelf. That's Distribution.
Shelf-Space also matters. Many people will walk into a store and if there's a huge block of a new game on the shelves, they'll buy it because well, it must be pretty special if it's getting that much shelf space. So when I walk into a store as a developer I'm thrilled to see 8 units of GalCiv II on the shelf. W00t! 
But next to it is Zoo Tycoon with 20 units on the shelf. Regardless of whether you like Zoo Tycoon or not, one must give credit to Microsoft for being able to push that many units into the channel and get the sales they get.
One of the things developers and publishers need to really explore is what role does "marketing" play. Do ads in magazines matter? What about webzine ads? There's a lot of research out there on how people find out about games. But there's little conclusive proof on that.
All we do know is that Distribution sets the potential and game quality X marketing determines what % of that potential you'll reach.