I recently ran a project on freelancer.com as an outsourcing experiment.

My motivation was mainly to learn a little about the process and its potential. I’m one of those people that knows a little bit of a lot of things and tries to do too much themselves. I wanted a minimal WordPress theme for this site but couldn’t find anything suitable. After flicking through the theme development guide I figured it would take me about 6 hours to create one from scratch. I have no interest in becoming a WordPress expert so this seemed a good project to outsource to one of the thousands WordPress developers out there.

I mocked up the design I wanted and set up quite a reasonable project description [1]. I deliberately avoided specifying implementation details and stuck to the top-level requirements. I wanted to approach this as if I were a business owner without much technical knowledge.

The noteworthy parts

Setting up a project brief and publishing was pretty seamless. I had to pay a few dollars by paypal and set a price range of $30-$250 (the minimum). Within 15 minutes I’d received a few generic responses and within 24 hours there were 12 bids.

Bid quality varied between three types:

Agencies – these responded quickly, most by automation. Their bids were generic, promoting their team’s skills and experiences. If they had read the project description they’d have known I wasn’t a good fit, although I imagine they’d do well using this approach as they all link to beautiful examples.
Independents that are poor at self promotion – these people sent one or two lines stating they could do the work and their price. They expected me to PM them for more information. I don’t understand their motivation, maybe it’s because they get harassed during this bidding process, but as they didn’t show any understanding of the project I wasn’t going to make more effort to learn about them.
Independents that get it – There weren’t many like this, but they were generally individuals that repeated back what they understood of the project, asked a good question or two and even estimated a timeframe for delivery. They provided enough information in a few lines for me to know the understood me.

The agencies charged significantly more than the individuals but, theoretically, you gain access to a project manager and their team. I think this could work for most people.

The good points

I selected a freelancer I liked immediately and he met my expectations. He asked good questions, he was enthusiastic, showed that he understood what I wanted and provided progress updates. He provided a first review and set the right expectations that there may be more iterations.
Upon completion I ask if he had any suggestions for next steps and he was open and enthusiastic about this. I think this is a good behaviour in a freelancer. I’d be happy to go back to him.

The bad points

The developer built the page exactly to my amateurish mock-up. I was hoping he’d have a little more creative insight to make it attractive. This was a poor assumption. I suspect it is better to be prescriptive about the creative design requirements.
Although he delivered exactly what was needed for the mock-up, a lot of best/standard-practice I would expect were left out. As a simple example, he set up some global CSS selectors for the menu list items that also applied to list items in the body of a post. That should have been obvious it wasn’t going to be okay, but the samples he demoed didn’t included bullets in a post, so everything appeared fine. I love the idea of interatively improving a system, but in the francelacer.com model the project is a one-shot. If I didn’t know how to fix things up myself, I’d have had to go back to him. Maybe that was the intent.

Lack of Ownership

This is an issue with outsourcing generally and is well described by game theory (agency theory). In a standard project, the freelancer’s goal is to achieve the project complete milestone with the least effort possible. They don’t bear the cost or risk of compromises made during the project. They have no ownership of the outcome, just the delivery.

On the other hand, the project owner has set a fixed price with little risk of cost overruns during the project, they have good control of the scope of the project, they control the milestones that the freelancer gets paid and they can apply reputation damage to the freelancer. This is great for the project owner, but the motivations aren’t well aligned.

Lessons Learned

Use freelancer.com for projects where the long-term outcome doesn’t matter. There’s plenty of short-lived projects that are a good fit. I’d use it again for simple, independent stuff.
Set up the project from the start where the freelancer has more ownership in the outcome. This could be an agreed maintenance milestone or some other benefit post-completion. I don’t think the suggestion of potential future work is enough. I’ll probably run another experiment based on this.
I’d like to run a similar experiment on oDesk. Their point of difference is they specialise in longer-running projects.

[1] The project was private but if you have a freelancer.com account you may be able to access it at project no. 1135018, “Create a minimal WordPress template for a blog”, posted by jeromyevans (me)