The Offering and Clients

Upwork is currently the largest freelance talent platform in the volume of work (i.e. the number of jobs completed). Launched in 1999, Elance originally focused on software helping companies manage their contractor workforce. It then evolved into a kind of job board for free-agent work that also facilitated payment and other services. oDesk launched in 2003, aimed at matching freelancers to projects. In 2014, Elance and oDesk merged, bringing their combined revenue to $941 million. As the Upwork website describes it: “Elance is where businesses go to find, hire, collaborate with and pay leading free agents from more than 180 countries… With a community of over 9.7 million free agents and 3.8 million businesses.” While the original Upwork clients were small businesses with limited in-house resources and a tight budget, more recent Upwork clients are Fortune 500 companies. The platform's popularity with big firms likely started blossoming when adventurous, tech-savvy managers within companies used Upwork for projects done on a limited budget and got good results. Then word spread, and over time, the idea of using Upwork became a normal, accepted option for any kind of work that can happen via the Internet.

The Upwork site includes a video from client Chris Clegg of Portland Marketing Analytics.³ He says when he started using oDesk, he would look for one free agent who possessed a mix of the skills needed for his project—say, a programmer who could do coding, design, and also know something about search engine optimization (SEO), or an entry-level analyst who knew Word, Excel, and PowerPoint. But this didn't turn out to be the most effective way to approach matters. Instead, Clegg says he learned it was better to deconstruct the work and hire specialists for each part.

Embracing the concept of deconstructing work and hiring specialists, like Clegg now does, shows how the talent platforms are not just a way to fill temporary jobs cheaply and easily. The real payoff of these platforms comes when you rethink the work itself, with an expanded perspective that doesn't assume you must use regular full-time employees, and that the work must be embedded within a job.

When Clegg limited himself to thinking about hiring a regular full-time employee for an existing job, he looked for someone who could do all the different tasks within that job. When you realize there are alternatives such as free agents and talent platforms, you can engage workers for elements of the job. You can get a designer to lay out the look of the site and choose fonts and images, engage a programmer to do the coding, then engage a search engine optimization specialist so that the site can be found.

The Work

At its core, Upwork is the Monster.com of freelance work. There are a vast number and a great variety of tasks offered there, and a large population of free agents available to do the work. The work that flows from Upwork is almost all done virtually. The most common types of work offered via the site are programming projects and creative tasks such as writing, graphic design, and website design. However, part of the beauty of Upwork is that it's big and diverse enough to find talent for almost any task. Upwork senior vice president Rich Pearson tells the story of a relative who dreaded composing a speech for a wedding. The solution turned out to be straightforward: Find someone on the talent platform who could write wedding speeches.

For programming work, there are two typical use scenarios. For projects involving new programming languages, leaders may have no in-house employees with relevant experience, so it makes sense to go to Upwork looking for that fresher skill set. The opposite use case occurs when your in-house programmers only want to do work on the newest platforms, and you won't be able to retain them if you keep them working on old languages. In this situation, the talent platform can be used to find free agents to do the work your in-house team considers mundane.

The process starts with a client posting a task and their budget (either fixed price or hourly). Free agents review the posted tasks and apply for work they would like. Clients can also search the database of available free agents and invite people to submit a proposal. The platform will also automatically recommend free agents who match the client's requirements. The system controls the number of people who can apply for a task; each post is limited to 30 invitations. This is an important control feature so that a client doesn't get hundreds of applications for one task, and free agents have a reasonable chance of winning the proposal.

The client then selects a free agent for the work in much the same way they would hire any other worker. They scrutinize resumes, portfolios, test scores, and reputation ratings to create a short list, then interview those on the list to make a decision.

The most straightforward type of project seen on Upwork is a small one, like the wedding-speech example, that can be done in a few days. However, clients also use the platform for large projects requiring multiple free agents over a period of months. Some go further, using Upwork for on-going staff augmentation where a free agent works perhaps 15 hours a week for an indefinite period of time.

The Workers

Put very simply, Upwork provides a place for free agents to find work, and this is key, as many free agents are good at their core skill—writing, designing, coding—but lacking when it comes to other parts of running a business, particularly sales and marketing. Talent platforms take away some of the need to market and sell yourself. The other major upside for free agents using talent platforms is the ability to work from anywhere in the world, at any time, for as long as one wants. There is a huge population of people who, for whatever reason, cannot or do not want to work nine to five at an office.

Free agents who use Upwork routinely make a couple of points about the experience. One is that you need to learn how to find work that is suitable for you and that pays decently. Navigating your way through the platform is not technically difficult, however efficiently navigating to the right work is a skill. Because many of the tasks are posted by small businesses that don't have much experience in that particular task, they may have unclear or unrealistic expectations. The free agent will have to be ready to deal with that. Free agents also say it can be hard to make a living entirely on Upwork. That is by no means universally true, and it may be even less true in the future if more work migrates to the platform.

The Engagement Model

Upwork provides a variety of tools to structure the communication between the two parties. “We are trying to build trust between the free agent and client,” says SVP Pearson.

The ways Upwork builds trust provide an interesting contrast to regular full-time employment. Yes, workers are paid, their work is evaluated, and there is a grievance system, but it typically takes place outside the employment relationship, often with far greater flexibility, quality, and lower cost. There are lessons here for you and your leaders, about how you could get these things done beyond employment.

Like employee performance management systems, there is an optional feature in which an Upwork tool takes random screen shots of the free agent's screen while they work on a client's project so that both the worker and the client can see when the worker is working on the client's task. There's also a free code review, so clients without technical expertise can be sure their programming work was professionally done.

Like an employee project management system that coordinates and tracks progress, Upwork has tools for organizing projects by milestones with ongoing status reports. To manage relevant files, there is a consolidated “work room,” a feature particularly useful for projects involving multiple free agents.

Like an employee sourcing and selection system, Upwork offers clients best practices, such as, “Do at least a brief interview by phone before you select a free agent” and, “Be very clear about your expectations, including whether the work is suitable for a novice, journeyman, or expert.” Like a recruitment and selection system for employees, the platform educates both clients and free agents in how to successfully deal with each other.

Like an employee grievance and dispute resolution system, Upwork monitors client feedback during the course of a project. If problems arise, they'll offer help and if necessary guide the client on how to begin a dispute-resolution process. If the dispute resolution isn't successful, Upwork will absorb the loss itself, paying the free agent for the work done, and letting the client use the funds to hire another free agent. This process is reassuring, however it is rarely needed; disputes occur in only 0.1 percent of projects.

Like an employee training system, free agents take tests on Upwork to get certifications that prove to clients they have the necessary skills, and to assess their own abilities and see what they need to learn. Upwork offers online training, some of it free. Free agents must take the initiative in learning, but Upwork makes appropriate training available.

Like an employee career and development system, the talent platform points free agents toward the kind of work they might want to pursue. The platform provides a transparent and real-time window into the demand for work, how well it pays, and what skills are required. They can match that to their skill to see how they fit into the competitive field for each type of work. The platform provides the tools, but the free agent does their own career planning.

Like an employee-reward system, the platform provides an escrow service that holds the money from the agreed budget for the work, releasing it according to an agreed-upon schedule, and on the approval of the client. Trying new things is also an important reward. In a traditional employment relationship, an accountant will seldom get the opportunity to try their hand as a speechwriter, but a talent platform gives them that chance.

For companies using a large number of free agents, Upwork provides a Private Talent Cloud that represents a central place for a company to control and track all work done via free agents. The clients can route all payments through the Private Talent Cloud, evaluate the reputation of free agents who have worked for them, track who's performing tasks for the organization, check who is currently available for other projects, and be alerted when coveted talent will become available. Upwork can even act as the employer of record for compliance purposes. The Private Talent Cloud is an example of a category of software called Freelance Management Systems (FMS). Others are Field Nation, OnForce, and Work Market.

If you think only of regular full-time employment as the way to accomplish sourcing, performance review, rewards, and dispute resolution, you miss these evolving new options. When these things are done through a platform, you can disperse the work beyond your organization, you can tap a pool of workers beyond those that are willing and available to be your employees, you can tailor your rewards to fit each worker and their contribution, and if your best worker leaves you have the arrangements in place to bring in a replacement without going through a hiring process. If using free agents carries practical difficulties, you can be sure that Upwork and the other talent platforms are working hard to overcome the problem.

What This Means for Your Future As a Leader of the Work: A Hybrid Model of Organization

Could Upwork itself be the model for the organization of the future? It's a hybrid firm of 225 employees and 500 free agents. Most of its employees do not have any direct reports, but each one is in fact a manager (and needs managerial skills) because they oversee free agents. A look at the organization chart of employees tells you little about what is going on since free agents outnumber employees two to one.

Upwork Pearson says the hybrid model is particularly useful for growing firms that are not ready to be locked into a static structure of jobs and employees. He personally uses free agents to get to those items lower down on his to-do list, the ones that otherwise might not ever get done.

It's fascinating, too, to see the birth and evolution of roles like “the free-agent wrangler.” These are free agents who help organizations pull together a team of other free agents for a project as more and more elements of an organization are being recreated within the talent-platform space. A free-agent wrangler is a manager who exists beyond the walls of an organization and beyond employment.

Of course, regular full-time employment does and will always have advantages, but the optimal way to get work done will increasingly blend platform-based work with regular employment, because the line between them is blurring. Your leadership framework needs to master that optimization.

The Offering and Clients

Tongal, a successful talent platform for creating video content, commercials, and similar products—perhaps eventually even feature films—is an excellent example of the new world of work, one unconstrained by the employment relationship. It delivers value to its clients, its talent community, and of course to Tongal itself. Its founders are James De Julio, Mark Burrell, and Rob Salvatore. De Julio's story starts right after college, on Wall Street, where he learned that he didn't fit in with that particular business culture. Instead of making a career there, he headed west to seek creative work in the movie business. He eventually landed a job at Paramount Pictures. De Julio noted that even though the business revolved entirely around creative talent (screenwriters, producers, cinematographers, etc.), the breadth of talent being considered for work was stunningly limited.

The industry had backed itself into a corner where everything was extraordinarily expensive; they would spend millions just to develop a story line. That was fine for a big blockbuster, but with the increasing fragmentation of the industry, this model was no longer working. People now are more likely to see what's available on Netflix than follow Hollywood's plan that they spend Friday and Saturday night at the cinema. Advertising faced a similar challenge: When you're preparing a commercial to be seen by millions sitting down to watch Modern Family, high costs are not a problem. But in the new world of audiences fragmented across multiple platforms—broadcast TV, cable, Hulu, Amazon—that cost structure would kill the business. De Julio says, “In 1986 you could reach 35 percent of the U.S. audience by advertising on a hit show; today no audience is like that. Adjusted for inflation it costs three times as much to reach an audience without even considering the fact they may be skipping your ads by using a DVR or tuning-out during your ad by doing something on their smartphone. The industry has transformed and models for creating advertising need to transform too.”

Tongal has attracted clients like Lego, Anheuser-Busch, and Microsoft. These firms are experienced in working with the best creative agencies, yet they use Tongal for some of their projects, presumably because Tongal does it better, faster, and cheaper. Yet it's no longer about cost cutting. Tongal has partnered with a movie production company to produce a full-length feature documentary. Nigel Sinclair, the production company's CEO and co-chairman, told the Los Angeles Times (September 17, 2013) that “Filmmaking is no longer elite. It's now a populist activity… What Tongal is doing is creating a way to find these new filmmakers and bring them across.” As online platforms like Hulu, Amazon, and Netflix start producing their own movies and television shows, can more such crowdsourced production be far behind?

The Work

The Tongal model is best understood in the context of the traditional ways of matching talent to work. A company that wants to make a TV commercial typically goes to its advertising agency, whose employees then brainstorm ideas for the commercial. After numerous meetings, presentations, and approvals, the agency begins to shoot the commercial, often drawing from its own small roster of free agents. This process is not all that different from a company having its own in-house creative department.

The Tongal process has three stages that use a familiar crowdsourcing contest model with a twist: First is the idea stage, where anyone from a creative agent to a consumer is encouraged to submit an idea for the commercial using no more than 140 characters (you can even submit your ideas as you fly across the United States on Virgin America airlines). The simplest way to do the next step is for the client to cull through the mass of ideas, choose the top three, and award the top three submitters a payment. The second phase is a video pitch stage. Using the three best ideas from the first stage, a new group of workers who can make a video pitch how they'd execute each chosen idea. Again, the top three are chosen, and all three win the cash needed to make a video. The third phase is when the best of those three videos receives the contract to make the commercial.

We state that the simplest way of choosing which ideas to develop is to have the client pick them. Tongal can also harness the power of its large community of creatives to assess ideas; and since it has demographic data on its community it can tell you not only which ideas the community thinks are best, but which ideas different demographic segments think are best. This “complication” is important because it shows what an amazing creative resource Tongal is sitting on, and how much more innovation is potentially possible.

The three phases Tongal uses in its process offer an important insight into creative work: There are many people who can come up with ideas but are not able to produce a video and many others with the skills to create a video who are not able to generate great ideas. It just makes sense to deconstruct the work so that idea creation and video production are separated. Tongal's entrepreneurial insight was understanding the work behind the industry and the creative process, and deconstructing and reconstructing that work into a process that can engage workers in ways other than employment. That new process looks very different from the typical studio approach, which structures the engagement as regular full-time employment or employment contracts. The Tongal process deconstructed the work and then reconstructed it into contests, iteration, idea formation, idea refinement, and execution. Thus, Tongal borrowed the work structures and tasks of a traditional employed manager in an advertising firm and morphed them to fit a world of nonemployees.

“We don't want to call it a marketplace; it's really a new alternative workforce model,” De Julio says. Yes, this is technically crowdsourcing, but it's not just a matter of casting a task or a problem onto the web and waiting for the crowd to provide the solution. Tongal shows that structure matters, and that structure reflects how Tongal figured out how to help clients lead through the work—not simply manage employees—by thinking beyond employment.

You can learn a lot by recognizing how Tongal deconstructed and reconstructed the work, reengineered its organization's boundaries to be more permeable, and adopted more customized and imaginative rewards. The same principles can help you to recognize when this kind of reengineering can produce breakthrough results for your work.

The Workers

There are a vast number of screenwriters in the world, but for any given project, in the traditional model, the studio only looked at a list of about 10 names. The same thing applied to other roles like producer or cinematographer. Why? Because when you approach the work through a traditional model of regular full-time employment, or even employment contracts, you tend to consider only those people whom you know well and who can perform all the elements of the defined “job.” The risk is simply too great to work with an unknown given that it costs on average $200 million just to market a new film globally, and there are few economical ways to evaluate talent before you hire them. This did not serve the clients for the work well, and it did not serve the workers well, many of whom were waiting tables while awaiting their chance to break into the industry.

Tongal has only 48 employees but sees itself as having 70,000 workers. That's a very different type of organization from the ones we've grown up with. Tongal cares about their workers, and while it structures the work, it looks nothing at all like a company with 70,000 employees.

These three profiles of actual Tongalers in the sidebar illustrate the types of workers that Tongal engages.

The Engagement Model

De Julio says “We have 70,000 people who have no obligation to do any work whatsoever.” They do the work because they are passionate about it.

Like an employee recruitment system, the Tongal contests are in essence a sales pitch. When you think about it, everyone makes a sales pitch to get a job or a work assignment. In many traditional employment systems, and particularly with creative media talent, that process can be very uncertain and time consuming for the worker, and it limits the number of candidates that an employer can consider. Tongal protects the interests of workers and clients by making sure the pitches don't demand an inappropriate amount of work. In the idea stage, the submission can be limited to 140 characters. That length is useful for clients who can now go through hundreds of entries in the time it might have taken them to meet just a few candidates in the traditional system. It also ensures that the workers (“creatives”) don't spend too much time on a pitch at the early stage when they are unlikely to win. A creative does not get into the heavy work of production until after they are awarded the contract.

Like an employee selection system, the Tongal second-stage pitch contest whittles a large number of applicants for the work down to three superior candidates. It also provides a focused and relevant set of information to help the client make the best choice among the top three worker candidates. Unlike employee selection all of the top three are “winners” and get funding to make a video.

Interestingly, Tongal even has a sort of “appeals” system if a worker thinks they were unfairly overlooked in the pitch stage. A “wildcard” creative can produce and submit a full commercial to the client if they want to invest the time and energy, and it will be considered along with the final product of the winning entry. It's a risky proposition for the free agent, who must produce the commercial at their own expense, but it provides a means to address the concerns of workers who feel they were unfairly passed over. The wildcard process has turned out to be quite important: one in six top prizes actually goes to a wildcard, to a creative person who was so sure they had the right stuff that they rolled the dice and made a video even though the client hadn't liked their pitch.

Like an employee pay process that aligns rewards with contribution, the Tongal talent community has the opportunity to earn a competitive wage. Forty of them earned more than $100,000 in 2014, and a great many more earn a substantial part of their income through the platform. Because its workers are not employees, Tongal knows that if the platform doesn't work for them, the best creatives won't come back. So, Tongal tracks the work and pay progress of its workers to make sure the system is getting the right work to the right workers at the right price. This is a common aspiration of employment reward processes, too, but traditional systems are often much less transparent and effective than Tongal.

Tongal also sets aside a pool of money so that workers who do well on multiple projects get a bonus. It is called Tongal Seasons, with a season being 20 eligible projects. For each project people finishing in the top five get points and at the end of each season the pool is split among the top 20 performers. It shows how some traditional reward elements can be brought into this distinctly nontraditional means of getting work done.

But the rewards don't stop there. Tongal imagines rewards beyond money. Similar to an employee engagement system, Tongal also has a big annual awards event: the Tongies. It's basically an award ceremony to recognize and celebrate the greatest accomplishments of the whole Tongal community. It blends competition, excitement, camaraderie, and mutual appreciation. Tongal flies in the top nominees from around the world, and the Tongal community sit nervously as they hear the words, “And the Tongie goes to…” If you thought of Tongal as merely a marketplace where free agents can pick up some creative work on the web, consider how different it is to be a Tongaler participating in the Tongies. Tongal is building organizational culture among workers who are not employees of any organization. Incidentally, Tongal is considering offering health insurance to Tongalers.

Like any employee total rewards system, the Tongal system has attractive and less attractive elements.

The Attractions

• The pay (which in some cases will exceed all but the very highest paid agency workers)
• The opportunity to build one's reel (samples of your work), which is often a precursor to getting work through traditional channels
• A chance to learn through competition
• The freedom to work on what you want, where you want, when you want
• The thrill of working with internationally known brands
• A chance to try directing
• Access to work that free agents would typically have little chance of landing because they don't have the relationships or live in the wrong location
• Protection from clients (Tongal sets out clear rules so that brands don't take advantage of creative talent; for example, by asking for multiple revisions of the film).
• A sense of community

The Challenges

• The lack of income security
• The expense of preparing for the competitions
• In most cases lower pay than what a director makes working for an agency on traditional television advertising
• Loss of money while working on a project for the sake of building one's reel and reputation

Notice how similar some of these attractions and challenges are to those of regular full-time employees of an ad agency or production studio. Notice as well that by organizing the work to fit a world beyond employment, Tongal can amplify many of the attractions. As a leader, you must ask yourself whether your system of regular full-time employment can compete in a talent market where Tongal and their peers are constantly honing their value proposition. If it cannot, then you need a framework to understand where you can dial up some things that Tongal does well (e.g., reconstructed work, imaginative rewards, community culture, etc.) without dialing down the elements that give employment an advantage (income security, benefits, etc.).

What This Means for Your Future as a Leader of the Work

At the moment, nearly half of all advertising spend is still on old-style TV commercials. As more and more of that budget shifts to digital and web-based platforms, the industry will be looking for more cost-effective approaches, and that will drive work to Tongal. Tongal is not limited to advertisements; the leaders' original inspiration was movies. One day we may see Tongal produce a feature film, it has already produced a winner of the Sundance Institute Short Film Challenge: the film, which you can find on Vimeo.com, is called 175 Grams, and it documents Ultimate Frisbee competitions in India. Tongal also has a partnership with Lionsgate on Stephenie Meyer's Twilight franchise. This project aims to create a series of short films based in the Twilight universe. Tongal has drawn on its vast community of creatives, which contains no small number of Twilight fans, to decide which characters to feature, and then goes on to create the films. The point is that Tongal is already spilling over the edges of the simple “we can make a commercial” model and using its potent talent community in creative ways. Perhaps one day a Tongal-like platform will assemble and coordinate talent for projects like designing a satellite, where assembling talent for a project has many of the same elements as assembling talent for a movie.

The future is a competitive game on an unstable landscape. For big-budget advertisements, traditional agencies with employees and contractors are currently the typical choice. In the world of Internet advertising, though, Tongal has the advantage, and agencies will scramble to adjust their model to stay competitive. Tongal will find ways to mimic the structures that give agencies their power; just as the agencies will be drawing lessons from the Tongal model. Talent marketplace evolution means that the less agile agencies and talent platforms will fail, as the agile ones find their unique niche—no doubt combining features of the old and new worlds.

The Clients and the Offering

Topcoder's clients are typically fairly large organizations and include famous names like NASA, Comcast, Honeywell, Salesforce.com, Harvard, and JDRF. The Topcoder model works best for larger organizations because there is an initially high cost to build the skills and processes to interface with the Topcoder talent platform. Once built, it's easy and not very costly to scale usage up and down.

While Topcoder case studies often lead with a statement like “project accomplished at half the usual price,” the real value is often superior quality rather than price. Their pitch is that Topcoder can get you a Google-like engineer at a fraction of the cost you would usually pay (presuming you could even convince them to work for you). For run-of-the-mill programming that doesn't require regularly and quickly scaling up and down resources, though, a traditional IT outsourcer is the more obvious choice.

The Work

Topcoder originally focused on only programming projects. The reason for this laser focus can help you understand the elements of work that make it highly suitable for talent platforms and the elements that may offer an advantage to other engagement models such as regular full-time employment. The project that works best for Topcoder, says Narinder Singh, chief strategy officer at Appirio (who acquired Topcoder), uses cloud-based computing. If the work requires a large application stored within the client's organization, like a traditional enterprise resource planning (ERP) system, it is much harder to deconstruct the work into pieces and disperse those pieces to the best talent, wherever and whenever they can do the work. When the work can be done with cloud-based applications, then the worker (usually a software coder) can access the tools and data they need from any location at any time without the client having to provide them special access. The work must be able to be deconstructed so that it can be dispersed. Singh observes that a single individual running an IT consultancy based on cloud-based applications can access the capabilities that previously might have required a 100-plus-person firm.

If you're the leader of one of those 100-employee organizations, understanding how Topcoder competes by systematically deconstructing and dispersing the work may be a key to your own survival. But there's more to understand.

The particular niche that Topcoder chose, software development and coding, also presented a unique talent pool that was already accustomed to a system of organization and rewards that suited cloud-based work design: programmers are accustomed to working on projects as contractors. Software and applications change so quickly that the clients for the work already realize they can't count on regular full-time employees to keep up. Both clients and workers had long experience using the cloud as the organization and location for the work.

As a leader, your job is to lead the work to achieve your mission. The Topcoder case shows how creative leaders saw a way to lead differently by deconstructing and dispersing work, creating rewards and an engagement model that fit the necessary niche of workers, and using the cloud as a more permeable and malleable organization. Does this work only for software programming? We get a hint of the broader future of talent platforms by looking at where Topcoder is heading: beyond narrow coding problems and into work in design, development, and data science. Having conquered one programming niche, they have begun to spread outward. As we will see in later chapters, you can also use these dimensions to more clearly see creative ways to lead through the work. You may not have a choice, as Topcoder may be heading your way.

The Workers

The Topcoder worker community is diverse and interacts with Topcoder in different ways. We dropped by the Museum Tavern in Toronto and ran into a couple of Topcoders. In a way, it is not too surprising. There are more than 700,000 Topcoders in the world. You could run into one almost anywhere. The ones we met at the tavern had studied programming, but their primary employment is bartending, a job they love that pays well. Participating in Topcoder from time to time is a way to keep their hand in programming and earn a little extra cash.

Another interesting case shared by Singh is about a programmer, who has a traditional full-time job in the Midwestern United States, yet earns 30 to 50 percent of his income on Topcoder. This is not secret moonlighting; his boss encourages him to do this because it's the only way to retain him. In fact, this programmer turned down a tasty job at Twitter because he found he had plenty of interesting work where he was and there was no professional need to move to San Francisco.

Some Topcoders are extremely active on the platform and use it as their sole source of income. Others use it simply as one source of free-agent work. Part of the strength of the platform is that it can appeal to a broad range of programmers.

The Engagement Model

Both Upwork and Tongal work in part because they can tap fresh and less expensive talent pools. In Topcoder's world, the economic driver is less about saving money and more about finding the very best person to do the work. Singh explains, “The most productive person in the world for a particular task is the person who has the skills required and who performed virtually the same task last week. The delta in productivity between that person and a smart engineer new to the problem is massive.”

How can you engage workers and work to take advantage of this massive productivity potential? If you think in terms of traditional organizations and regular full-time employment, you look within your organization and choose the best person out of 10 or 100 or perhaps 500 employees, and hope they are available. By deconstructing and dispersing work to the cloud, and engaging nonemployees, Topcoder can use a pool of hundreds of thousands of programmers. Remember Ion Torrent's need for a star programmer who worked on a particularly tough data-compression problem last week? The likelihood that this person sits among the employees of any one organization are remote. The chance they exist somewhere in the Topcoder universe is much higher. Further, because you can focus that person on the task they are best suited for, their performance advantage is magnified.

Like a piecework pay system for employees, Topcoder does not pay for time worked, but rather for a unit of output. The pay-by-unit-of-output model is clean and easy to manage; however, it presumes you can deconstruct the work into appropriate chunks. Singh says that deciphering that enigma is one of the critical capabilities that Topcoder brings to projects. Indeed, the inability to properly deconstruct work into meaningful activities is often the downfall of traditional piecework systems applied to traditional employees. A significant element of value to clients is that Topcoder helps them deconstruct and reconstruct a project into parts that can be accomplished and rewarded via the talent platform. The step of breaking apart the work can be handled in various ways. Companies can have one of their own IT employees act as the disintegrator and integrator. Another approach is to use the Topcoder platform to tap its worker community for the best copilot to assist with the project.

Like an employee learning and development system, the Topcoder platform offers a powerful way for programmers to develop their skills. With thousands of software projects available to the community, Topcoder allows programmers to do real work on interesting problems and have it reviewed by peers. It's that side of Topcoder that keeps bartender-programmers engaged. In a traditional job, that bartender might have only a small fraction of the experiences they can have on Topcoder.

Like a performance evaluation process for employees, Topcoder workers get visibility based on the quality of their work. That work quality is measured very precisely, because Topcoder deconstructed the work to make the activities, output, and their quality highly visible. Coders' output can be evaluated not only for its functionality, but also for whether the code itself is elegant. Coders are rated red, yellow, and blue. Those with a red rating can step to the front of the line for corporate jobs. The yellow- and blue-rated programmers are also very skilled, and that rating helps them get good work. In a cloud-based system, your reputation carries far more weight than a performance rating in a single employer's database.

Similar to a culture-building celebration for employees, Topcoder holds the Topcoder Open, an annual gathering that brings 100 top programmers from around the world to participate in a tournament. Videos at the Topcoder website show a participant who said, “It's a chance to see the very best people solve some brutally hard problems really fast.” The passion and enthusiasm of computer coders to show who's the best rivals gatherings of spectators to watch stars play interactive web games like League of Legends. Talent-platform workers do not share an office or a company logo; however, they are a community with a culture. There is more to life than CEO of Me and no doubt in the midst of an event like the Topcoder Open, the contestants will describe themselves first and foremost as Topcoders.

Why would Topcoder have any regular full-time employees? The engagement model for their employees provides you with a clue as to how best to optimize the combination of work done through employment and work done beyond employment. An employee handling demanding projects may have no employees reporting to him but have dozens of people working for him—all free agents engaging through the Topcoder platform. On the typical organization chart, this employee looks like an individual contributor with no subordinates. In reality, his span of authority is more similar to a high-level manager or director. His value is a unique ability to take the work that clients bring to Topcoder and deconstruct and reconstruct it to fit the population of workers that he can tap in the cloud. That kind of unique capability resides in an employee of Topcoder, because it cannot easily be deconstructed and farmed out to the cloud and because Topcoder leaders anticipate a steady need for that capability across virtually every cloud-based project they receive. Appirio has 1,000 employees whose value is multiplied severalfold by the company's capability to engage cloud-based workers and cloud-based work in ways that are beyond employment. The company can have a lean internal structure, with all the benefits of the simplicity that come from having a manageable number of employees, but still punch far above its weight because it has the know-how to tap the skills of thousands of free-agent workers. Even here, for common projects like developing a mobile app, Topcoder can set out standard guidelines on breaking the work apart. It seems inevitable that one day Topcoder will have “wizards,” interactive programs that will guide an IT manager on how to parse the work into various common projects.

What This Means for Your Future as a Leader of the Work

Can you bring the Topcoder model in-house to your regular full-time employees? Some CIOs use Topcoder to peer into the quality of their employees. Traditional employment systems often cannot tell a CIO if the best team of Java programmers is in their Singapore office or in Poland, or laboring away in obscurity in Milwaukee. A Topcoder contest for employees provides an X-ray for assessing regular employee talent. Topcoder already has benchmarks and programming challenges developed by analyzing millions of coding assignments and outcomes, so why try to invent them yourself? As a bonus, the Topcoder database already spans thousands of workers, so you can see how your internal staff matches up to the rest of the world.

As a leader, should you use contests to measure and find the skills among your employees? It's a good example of a hybrid model that takes one of the best elements from the world beyond employment, and combines it with the world of employment. Yet, if you suggest to most HR departments that you're going to pay your employees through contests, they will correctly tell you that this might strain your existing employment systems to the breaking point. Because your systems are based on a traditional model of regular full-time employment, you'll run into the inertia of compliance, established HR processes, and deeply rooted notions of the meaning of pay equity. As a leader, your most important task may be to oppose that inertia, or to hire HR leaders that are prepared to break tradition.

NASA is working with Topcoder to run competitions called Single Round Matches (SRMs) to assess their employees' skills. Topcoder has good benchmarks on SRM performance, such as how many people solved a particular problem, how well they solved it, and how quickly; these benchmarks allow NASA to get a great handle on how good their internal programming team is. One large bank is even using this benchmarked internal-only Topcoder competition to allow their strongest performing employees to show off their skills and pick their “dream job” based on winning.

Topcoder is continuing to hone its model. For example, leaders there looked at whether it's best to let programmers work on problems independently or to get them to share ideas as they go. Topcoder found that having programmers work independently creates a great diversity of ideas, while sharing ideas enhances collective learning. This gives the company a process for structuring work, starting with independent work to create a diversity of ideas, hitting a checkpoint where leaders bring people together to share and create convergence, and then opening it up again to independent work. This sort of sophisticated process improvement makes Topcoder and platforms like it ever more effective alternatives to traditional methods of getting work done.

The Offering and the Clients

Some conundrums, such as identifying duplicate web pages, are simple for humans, but too hard for computers. Amazon created a platform where this kind of work could be broken up into tiny pieces and distributed to a low-cost workforce. Amazon calls it “artificial intelligence.” One feeds a problem into the MTurk engine and back comes an answer; behind the scene are humans, nearly invisible to the user.

MTurk has a wide range of clients including AOL, the U.S. Army Research Laboratory, Statera (an Internet media metrics company), and university researchers. It takes a certain amount of sophistication for a client to use MTurk, but it's a natural tool for anyone handing large amounts of data.

Amazon liked MTurk so much that its leaders launched it as a subsidiary in 2005, and it has grown into a sizeable business with access to half a million workers in almost 200 countries.

The Work

The work generally requires no special skills, and clients really only care to distinguish between people who do work of adequate quality and those who do not. There may be significant skill differences in various tasks. For example, some workers may be much faster than others in transcribing grocery receipts. However, that skill difference doesn't matter to the client who is paying a piece rate.

Some examples of work to be found on MTurk include

• Record a word or phrase in your own voice (pays $0.02).
• Transcribe all of the purchased items and total from a shopping receipt ($0.09).
• Find the headquarters of a particular company ($0.10).

Typically, there are thousands of cases for each task, so a Turker (as the workers are called) might do 500 cases of “recording a word or phrase” and earn 500 × $0.02 for a total of $10.00. Amazon charges the client 10 percent of the amount earned.

Amazon calls the units of work Human Intelligence Tasks (HITs). These are microtasks, far smaller units of work than most managers are used to seeing. Microtasks are quite a common feature in the information age. Main categories of work are

• Photo and video processing (such as tagging images)
• Data verification and clean-up (such as verification of a store's hours of operation)
• Information gathering (such as filling in a survey)
• Data processing (such as podcast editing)

This work can be done virtually and usually requires no special skills.

Tagging photos is the classic example of a task that is easy for people and hard for computers. That is changing. Research by University of Toronto assistant professor Ruslan Salakhutdinov and others on “deep learning” have, in the past few years, led to stunning advances in creating algorithms to enable computers to accurately tag photos. Ironically those smart computers are being trained, as you probably have guessed, by Turkers.

The Workers

Panos Ipeirotis, a New York University business professor, has researched the demographics of the MTurk talent community. In a 2010 survey he found that 46.8 percent of MTurk's labor force was American, 34 percent was Indian, and 19.2 was “other.”⁴ The workforce will likely become increasingly global as more people discover the platform. In the United States, Turkers are often stay-at-home parents and younger unemployed and underemployed workers, with women Turkers about twice as common as men. In India, male Turkers are about twice as common as females, and the platform is more likely to be an important or even sole source of income. Even though HITs are low-skill work, most Turkers have a university education, which is likely reflective of the fact that educated, computer-literate people are most likely to discover MTurk. Average earnings are about $2 per hour, according to Ipeirotis. For developed-country workers the platform can serves as a way to earn a little extra money using time that might otherwise be spent surfing Facebook or watching videos.

The Engagement Model

The MTurk model is far different from platforms like Tongal, Upwork and Topcoder. It relies far more on the contractual relationship between the client and the worker, providing far fewer connections between MTurkers as a community and far less infrastructure in support of its workers. That's not necessarily a bad thing, as both clients and workers are fully aware of the arrangements when they undertake the engagement.

There are several ways clients can be sure the Turkers taking up a task are qualified to do good work. One way is to require Turkers to take a test before being allowed to work on the task. The test becomes the first task the Turker does for the client, and if they pass, they can move on. Another method is for the client to restrict their work only to experienced Turkers. A Turker's forum noted that to get most of the better-paying requesters (i.e., clients) requires between 5,000 and 10,000 approved HITs to be completed before you can work on theirs. Perhaps the most mechanical way to insure quality is to give multiple workers the same task. If two workers give the same answer (e.g. add up a receipt to the same total), then the answer can be deemed to be correct. This last method is particularly suited to a platform that can quickly access workers to do standard tasks at low cost. A traditional employment model probably carries too much overhead to consider giving two employees the same task very often.

Unlike Topcoder, the MTurk talent isn't celebrated in contests on YouTube and isn't visible to most clients. While Upwork has a dispute-resolution mechanism, MTurk clients simply don't pay for work they don't like, and are not required to give a reason. In the absence of the sort of loving care Tongal provides its community, third-party groups have sprung up to support Turkers such as Turkoptican, which can alert workers to shady employers, and CloudMeBaby, which provides a forum for people using MTurk and other platforms.

What This Means for Your Future as a Leader of the Work

MTurk illustrates how talent platforms can span a wide range of work and workers, in this case creating a model uniquely honed to work that can not only be deconstructed into small parts but can be done with widely available capabilities. It also illustrates that talent platforms need not be devoted to only highly qualified talent where the objective is to find the best and pay them highly. Indeed, many writers lament the possibility that a world beyond employment will look more like MTurk and less like Topcoder or Tongal, with a race to exploit workers as commodities.

As a leader, this diversity has an important message for you. Your future task of leading through the work will require that you segment the work and the workers at least as precisely as you segment your customers and your product offerings. Regular full-time employment is only one option, but even when you recognize the possibility of using talent platforms, the diversity is stunning. MTurk shows the diversity of niches that are open to talent platforms, and that diversity will only grow, creating more opportunities for leaders to tap free agent world.

The changes in the way we work and the way we buy appear to be moving quickly, but pull back and you'll see that such evolutions do take time, even when the Internet is involved. As Gary Swart, former CEO of oDesk points out, after 20 years of ecommerce, still only 7 percent of all sales take place online. We started by buying books on the web and then progressively moved to electronics, financial products, cars, and even groceries. But even so, this accounts for only a small percentage of current consumer spend.

Similarly, companies will dip their toes in this new space by segmenting the work and testing the effectiveness of these platforms. There will be early adopters who push many different types of work onto these platforms, while others will sit back and watch. As we move up the maturity curve, we'll get more comfortable as we better understand the nuances of work and how best to get it done. Talent platforms are like the first fish crawling out of the sea hundreds of millions of years ago. They do not look that impressive right now, but what will they look like with time and evolution? Will they become giants striding across the earth, filling a thousand ecological niches?