It can be an appealing idea to go beyond your regular community of expertise to meet a challenge, and there is a growing interest in what can be achieved through crowdsourcing. But there is a patchy understanding of what makes it work successfully.

"The bigger misunderstanding is believing that we can harness a crowd," says Stefan Haefliger, reader in strategic management and information at the Centre for New Technologies, Innovation and Enterprise at Cass Business School. "Sometimes this is deeply problematic, because the individuals providing solutions are not being harnessed, they are doing what they want."

Just casting the net widely for ideas is unlikely to produce solutions to a specific challenge. Instead, a successful crowdsourcing project is likely to depend on three main elements: the right crowd, the right incentives to take part, and the right question.

Haefliger makes the point that "it has to be faced towards the type of crowd you want to engage", and Alph Bingham, founder and director of crowdsourcing specialist company Innocentive, says: "You first have to ask what type of partners you need, how you can gain access to the minds that possess the information you're looking for."

There are often ways into this through connections with customers, social networks or academic communities, and incentives to take part vary between groups. Haefliger says 'pre-emptive generosity' can be productive, giving freebies to people who could be interested related issues to create goodwill. This has been used in software development and companies in the entertainment industry are trying the approach.

Competitions can also stir up interest. Crystal manufacturer Swarovski has run competitions for designers that give the winners an upmarket sales outlet while it obtains new talent for its business.

Bingham says that prizes come behind the more intrinsic motivations of meeting an intellectual challenge to prove personal credentials. But he believes that if it is a difficult challenge a cash reward can show that the company really values the responses.

"People are also using the cash as a score keeping mechanism, a way of articulating how much value you place on a solution," he says. "In that sense it doesn't show up as an incentive, but it is an important part of the platform when engaging a crowd."

The big test comes with the formulation of the question to get the best possible solution. Bingham emphasises that this can vary between appealing to people in one industry or a broader community.

"If I formulate something and clearly describe it as an automative problem, I have a bias towards the automobile engineers," he says. "If it's do with physics and material science I want to keep it in as flat and open a form as possible."

He adds that when faced with a big challenge it's often best to break it down, and maybe to move some elements away from crowdsourcing.

"Best practice would be to take the large complex problems and break them into individual modules," he says." Just because it's a portable module doesn't mean you have to port it off somewhere. It could be that work on that module is highly confidential and therefore needs to be done internally, or the work is highly specific to your organisation and needs to be done internally.

"Nevertheless, having broken down your big problem, you can say the appropriate way for dealing with each module is to contract it to a specialist lab, or to crowdsource it, or to partner with an academic institution. You've got a lot of choices."

Crowdsourcing is being subjected to some major tests. Last year the UN High Commission for Refugees launched its UNHCR Ideas programme with a challenge to make it easier for urban refugees to confirm their status and receive support without travelling long distances. This was open to agency staff, their partners and refugees.

It used the Spigit Engage platform, developed by software firm Mindjet, to collect more than 150 ideas over five weeks. Ten were reviewed by an executive committee and one was chosen for implementation – a website named that enables people to register through an internet terminal or mobile phone. The next challenge, aimed at helping refugees to learn a new language, is due to run this year.

Olivier Delarue, the UNHCR lead for innovation, says a couple of strong lessons emerged from the first project. One is that the ability to float an idea anonymously on the platform, even though the organisers know who it came from, can encourage some people to take part.

"Then the wording of a challenge, if it is not engaging enough, may not be enough to generate the ideas you were hoping for," he says." This is a new science, crowd activation, not just about crafting the right messages but continuing the engagement process of staff."

While the UNHCR does not give out cash prizes, it does give staff points for the best ideas, and these are included in appraisals and can be a factor in promotions. Their names can also go onto the platform, providing peer recognition as innovators, and there is a fund to develop the best ideas.

The recognition can also work for refugees, with the possibility of support from the UNHCR programme for entrepreneurship development, and of being hired by companies in relevant fields.

In fact, the inclusion of refugees in the process highlights the prime feature of crowdsourcing - it draws on the ideas of people who would not usually be in the loop.

Alph Bingham sums it up: "It's very important because the problems are rarely solved by people with a resumé that you would have hired to solve that problem. They are usually solved by someone peripheral to the field who has seen that problem in a new light."