How to sell software?

Sales models

Fundamentally there are two ways to pay for software: either the end user (or somebody on her behalf) pays, or someone else. The second option is usually called free software where either advertisers pay (and the user pays by having to live with the ads) or a sponsor (which may also offering not money but work). 

Neither solution is optimal from a business perspective. There are people who are willing to pay a lot for a software and a lot of those who are willing to use it but don't want to pay.  The diagram shows the relation  between price and number of users. The vertical axis shows the price while the horizontal axis the number of users. The points of the diagram shows the price and the number of users who accept that price. Nobody is willing to pay above a certain price and there are those who are no willing to pay anything but are ready to use your software.

For the developer both groups are valuable; the first for obvious reasons the second because they also deliver valuable feedback and may evolve to paying users once.   If the application is sponsored even then there is a price - number of users relationship. 

Feature auction

People use a program for it's features (functions, usability, speed...).  What if people could select which feature they want and how important it's for them. That is what I would call feature auction. You make a list of features and people can offer a money for the feature which is most important them. The feature which receives the largest amount of donation will be implemented first. You can keep control by selecting the features you think is important and your users the one which is most important for them.

Problems

• Losers loose they money. It shouldn't be the case, it can remain on their account (even "paying" some "interest") and can be used for the next bet.
• No interest. Yes it may happen to any new software. It that case you may also receive "free votes" which can't compete with "paying votes", or simple decide at your wish. Either way you must build trust in your users, otherwise they can be disappointed very fast and cease supporting your model.
• Predictability. Businesses don't like the unpredictability; if their pay for something they must exactly know what they get for it (which is not so common in real world). A donation which either gives them what day want or not is not a good investment.  You can ease this problem (e.g. with votes before betting, send money back - which is painful, provide some guarantees for subscription...) or offer a fix price for a feature.  Neither of it is optimal. 
• Contradictions. You may have contradicting requests. In that case you should consider to start an applications family with identical core and different feature set.
• Free rides. You may have to many free riders, who know that you will make the most important developments anyway. Maybe you have to offer them a lower level product.