It talks about retail, where the last mile, getting your good or service to the consumer:
accounts for the bulk of the cost of the service or good being sold.
Retailers get half. Insurance people get commissions. Distributors make their share.
In the downloadable distribution model of games, we run into the same issue. After making the game, how do you get it to consumers? Though it seems like the Internet should solve this issue and make it easy to find people to buy, the sheer amount of information on the Internet makes finding customers as difficult as it was in retail. (which, btw, is why casual game portals can give developers a 25-40% of the revenue and keep 60-75% for themselves)
Seth addressed this point and should get you thinking about a solution:
If you're not going to plan on paying the messenger, your offering better be so remarkable and have such a viral story that your investment in product eliminates the need for media and sales.
Many Indie game developers start out with the idea that they can simply make their game and that people will naturally pass it on. Viral marketing is really hard. There are very few products that are viral. Expecting your product to be viral is probably not the best approach.
What is the best approach?
That depends on you and your product. At the Game Developer's Conference in March of this year I did a session on marketing Indie games. I suggested that developers should use the portals, and abuse the portals by placing things in their games such as level editors and obvious, but obscured secrets, that would cause players to go out looking for the developer's own website. Over time, you can build an audience that comes directly to you.