Deloitte University's Industrialized Crowdsourcing is one of the best, relatively short descriptions of crowdsourcing I've read.
The key quote is the summary:
Enterprise adoption of the power of the crowd allows specialized skills to be dynamically sourced from anyone, anywhere, and only as needed. Companies can use the collective knowledge of the masses to help with tasks from data entry and coding to advanced analytics and product development. The potential for disruptive impact on cost alone likely makes early experimentation worthwhile, but there are also broader implications for innovation in the enterprise.
The article goes on to describe the major applications of crowdsourcing:
1. Simple, Task Oriented Crowdsourcing: Shot, transactional units of work where the workloads can split across large numbers of remote workers. Examples include data entry and picture tagging. Also includes routine tasks requiring physical presence such as performing store pricing checks.
2. Complex, Experienced Based Crowdsourcing: Work that requires abstract thinking, specialized skill sets, and sophisticated problem solving. Examples includes solving scientific or business problems.
3. Open Ended, Idea Generating Crowdsourcing: These usually involve challenges oriented around invention, idea generation, and product and brand innovation. Examples include crowdsourced TV commercials.
4. Crowdfunding and other Crowd Contribution Crowdsourcing: In these types of crowdsourcing the crowd contributes something of value. Examples include crowdfunding as well as the contribution of information or other items of value.
The article goes into more detail and has lists of companies providing crowdsourcing services in each of the categories.
With the exception of crowdfunding, most users of crowdsourcing services are larger firms. But use is spreading to smaller firms as they realize the benefits of crowdsourced work - low costs, speed, scale - also appy to small business tasks.