Skip to Content

Crowds of Workers, on Demand

Companies are turning to crowdsourcing middlemen to make “good” crowds.

In 2009, Kraft Foods wanted a catchy name for its new version of Vegemite, a yeast-based spread popular in Australia. Instead of hiring a branding agency, the company decided to crowdsource the task. Kraft asked people in Australia to submit and vote on potential names for the new product. After 48,000 entrants made suggestions, the winning name was “iSnack2.0.” When the company announced it, bloggers and fans ridiculed it and the company retracted it a few days later. Not only was the name scorned, but it had also been trademarked previously.

Ranking system: This screenshot shows how various Trada contributors stack up in the work they’ve done.

The iSnack2.0 incident illustrates the perils for businesses that try to take advantage of the supposed wisdom of the crowd. It’s no mystery why the idea appeals: instead of hiring one person for a task, a business can pay little or nothing to divide it up among thousands, often getting the work done faster to boot. But while companies have had some success divvying up simple jobs like categorizing products, harnessing the crowd for more complicated tasks has proved difficult. Even as crowdsourcing became a buzzword, it became apparent that getting useful input from a faceless mob in an unstructured online environment was harder than it seemed.

Now, third-party companies have sprung up to act as liaisons between businesses and diffuse groups of potential contributors. These companies say they can attract specialized crowds to help solve a wide variety of problems in such diverse industries as chemicals, design, and the Web. “It turns out that when you have tasks that require creativity and planning at a higher level, the overhead involved and the need for consistency across the whole task makes [crowdsourcing] very difficult,” says Judd Antin, a research scientist at Yahoo who studies online collaboration. “We’re at the beginning of understanding how to take advantage of the efforts of many of thousands of people.”

To start off, one needs the right crowd. That’s apparent from the company Threadless, often cited as a successful example of crowdsourcing. Threadless, which sells T-shirts online, asks people to submit ideas and vote on designs that they like and would buy. The Threadless staff assesses each shirt for ranking, comments, and “social score” (the number of tweets about a design, or the number of times people logged in to Facebook have clicked that they “like” the item) and picks a final design to print. The power of Threadless lies in its crowd: 1.5 million voting members have submitted over 300,000 T-shirt designs.

But few companies have such an engaged community or time to build one. That’s where crowdsourcing companies like CrowdSpring, CrowdFlower, Victors & Spoils, NineSigma, and InnoCentive come in: they offer crowds for specific tasks.

Trada, founded two and a half years ago, straddles the line between traditional outsourcing and crowdsourcing: it takes on a company’s project and farms it out to a widespread pool of people. In particular, Trada gets these people to help companies create and improve ad campaigns on search engines like Google, Yahoo, and Bing. A tourism company, for example, would bid on keyword combinations (like “Hawaii vacation” or “Europe tour”) and write short text for an ad. When a Web surfer types in that keyword, the company’s ad comes up alongside the search results, and if the ad is clicked on, the company pays the search engine the amount it bid. The ways of improving an ad campaign are numerous; companies can switch or combine keywords or make them show up only at certain times of day, for instance. But comparing results and making the changes can be laborious. Those are the tasks that Trada turns over to a crowd of workers. If a business aims to spend $1 a click and a Trada worker creates a successful campaign for 80 cents a click, Trada and the workers keep the difference. Trada’s customers include Warner Bros. Music and the New England Patriots.

Trada has more than 1,500 people available for such work; each ad campaign can require five to 20 contributors. After passing a basic certification test, a worker has to accomplish three out of five goals for a customer before gaining access to more campaigns. One such worker is Jeff Yin, who was a laid-off climate scientist when a friend recommended that he sign up with Trada in 2009. He’s currently working on about 30 campaigns and makes about $4,000 a month working 30 hours a week. “It’s a way I can learn about Internet business and get great experience in pay-per-click advertising,” says Yin, who hopes to pursue his own Internet business someday.

In addition to attracting the right crowd, a crowdsourcing platform has to provide incentives to do what a company wants people to do, says Niel Robertson, Trada’s CEO. Aside from offering monetary compensation, Trada tries to develop a strong sense of community, making use of game elements to keep workers interested. They can earn points that bring them into “levels” and get public displays of their reputation ranking. “You can never tell a group what you want them to do; you have to incent them,” Robertson says. “I think people are surprised by how much curation goes into it.”

Building reputations and discouraging anonymity also help promote successful collaboration, says Robertson. In Trada, no one is allowed to work anonymously. “If you run a public process that’s not anonymous and curate the crowd, good content builds on top of itself,” he says.

While Trada builds a specialized crowd, other crowdsourcing providers are taking a different route: they offer to create custom-made crowds. One such service provider is called Chaordix. “Crowd recruitment shouldn’t be underestimated,” says Randy Corke, the company’s vice president of business development. “We create different crowds for each client [and] can implement different models.” Some models encourage collaboration between participants in a crowd, for example, while other models keep the crowdsourced work secret.

When the University of Oxford asked for help brainstorming ways to help reduce maternal mortality in developing countries, Chaordix needed to find doctors, nurses, and midwives in those countries. It helped find medical and research organizations outside Oxford to reach out to. The company and the university also encouraged Oxford medical students going on internships in developing countries to hand out surveys and invitations to participate in the crowd. Chaordix also mines websites and uses social-media tools like Facebook and Twitter to recruit people.

“In many cases the companies we work with already have the seeds of their crowd,” says Corke. For instance, companies might have lists of potential participants in newsletter lists or customer loyalty programs. “If they don’t, then we work with them to determine the characteristics of people we should invite into the crowd,” he adds. People possessing a certain skill or knowledge, for example, or a particular demographic might be useful, but Corke says Chaordix also looks for diversity—geographic or otherwise. “Generally speaking with crowdsourcing,” he says, “the more diverse the crowd, the stronger the results.”

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

The problem with plug-in hybrids? Their drivers.

Plug-in hybrids are often sold as a transition to EVs, but new data from Europe shows we’re still underestimating the emissions they produce.

Google DeepMind’s new generative model makes Super Mario–like games from scratch

Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.

How scientists traced a mysterious covid case back to six toilets

When wastewater surveillance turns into a hunt for a single infected individual, the ethics get tricky.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.