Is the Best Career Erratic or Predictable? Data from Freelance Platforms Offers Clues

If you were hiring an actor for a romantic comedy, would you prefer Hugh Grant, whose roles are consistently the leading love interest? Or Arnold Schwarzenegger, whose resume jumps from The Terminator to Kindergarten Cop to Governor of California? Or Tom Hanks, who has had a systematic progression from television to romantic comedies to more serious dramatic roles?

Hollywood may be an extreme example, but this is the kind of dilemma today's leaders are faced with. In the emerging work ecosystem, a life-long career and "perfect fit candidate" is a thing of the past. Today, people have a plethora of work options—full-time jobs, projects, freelance gigs, contract work or part-time employment—which often means they have erratic job histories.

This new world of work opens a pressing question for organizations: Is a modern candidate more attractive if they've held a single similar position, progressed through a closely-related sequence of positions, or moved with agility between very different positions? While the answer isn't certain, data from freelance platforms offers a few clues.

When careers occur inside one or a few organizations, candidates tend to move systematically from entry-level positions to leading small teams to leading departments. But in today's world, careers are no longer constrained by such traditional employment rules.

Like Arnold Schwarzenegger, candidates can more easily jump between very different areas of focus, arguably demonstrating their ability to be agile, learn quickly and adapt. Freelance platforms enable thousands of people to find such diverse work experiences every day: For example, Upwork—a freelance platform—reports $1 billion in work done annually, with more than 2,700 skills available and four million clients.

But while freelancers may be pioneering a new career path with such platforms, are those who hire freelancers equally open to job erraticism?

Social scientists are increasingly studying work through freelance platforms. Ming D. Leung, a professor at University of California-Berkeley's Haas School of Business looked at the question of career erraticism on freelance platform eLance (now merged into Upwork) in 2014, examining 268,000 bids for 20,000 jobs from 2,400 qualified bidders.

Pairs of jobs were deemed similar the more often they appeared together in a candidate's work history. The diagram below shows the job clusters Leung discovered. For example, the left-center represents the web and programming domain, where jobs are quite tightly clustered, while the bottom-right shows the legal domain which has less clustering.

A candidate's past job sequence was used to calculate the average distance between the consecutive jobs they worked. For example, jumping from "3D graphics" to "tax planning" garners a high distance score compared to the low distance of jumping from "3D graphics" to "animation." The overall erraticism score was then calculated on a range between zero and 1.0, with an average of .41.

How did the erraticism score impact a candidate's chance at getting a job? Movement between similar jobs is better than staying with one job, but only up to an erraticism score of .18. Beyond that, more erraticism reduces the chance of winning the job.

However, some factors reduce the negative effect of erraticism. Erratic job hopping had less of an effect on simple jobs, freelancers with a longer job history on the site and freelancers who had prior contact with the employer.

The freelance world and the traditional world share similarities: Freelancer movement between similar jobs is still more attractive than erratic movement, and proven experience and a past relationship make erraticism more tolerable.

Will this change as employers of freelancers learn more about the patterns that lead to success? One thing seems certain: Freelance platforms will increasingly be a rich laboratory to observe the evolution of work, and leaders should pay close attention as we learn more.

Photo: Shutterstock