Labour marketplaces are always in the news - earlier for democratising access to skilled workers and now for the ripple effect they create in the society and the need for universal basic income. Irrespective of which direction the wind flows, they are some of the hardest businesses to build and scale.
I admire Upwork and Elance, because they made freelancing a mainstream option for several thousand workers. They made the 'impulse' purchase of skills friction-free. Got a Wordpress blog plugin to upgrade? With USD 10 and 30 minutes, someone from Bangladesh can be at your service.
There are many marketplaces that are unbundling: Upwork, Elance, Freelancer etc. Gigster is attempting to identify the cream of developers. At ContractIQ, we are attempting to aggregate dev shops for mobile and web product development.
Broadly, they all do the same three things, in pretty much the same way:
Of the three, most labour marketplaces will solve the first problem of discovery, very well. The problems of evaluation and fulfillment will be solved with varied degrees of reliability.
The 'evaluation' problem
Why is it hard to solve the 'evaluation' problem?
There are two reasons:
Let's say you are looking for a PHP developer and you've got someone with four star rating across 30 projects. That's statistically significant on the surface. However, if all their works have been for small fixes of less than 40 hours each and yours is a multi-tenant application for the enterprise, the statistical significance has no value. In simpler words, you are looking at the wrong guy, who just seems like the right guy.
Today's feedback systems are too shallow and rigid that you cannot apply your context and derive meaningful inference from data.
If this is what you have on your hands to shortlist providers, you may end up picking the wrong provider. For the economists among the readers, we are talking about 'adverse selection' here.
Online marketplaces are built for the lowest common denominator usage. You can compare providers by price, location, language, reviews etc., but not beyond that. For example, if you need a service provider with the understanding of how credit rating systems work, so that they can build the workflow for a better lending platform, you're out of luck now. Even the best of machine learning and natural language techniques cannot solve the problem of contextual match-making today (or at least to an extent that makes it meaningful).
Unfortunately, customers do not care if you meet their least expectation (which is what filters and faceted search are for). They want exact solution to their problem. For non-e-commerce, service-aggregation marketplaces finding the exact match is a problem that's yet to be solved.
The 'fulfillment' problem
Unless your freelancers are completely dependant on the platform to generate their livelihood, they are not hard-pressed to keep you happy. The marketplace is one of their sourcing channels and if there is bad feedback there, they go to competing marketplaces or find work offline. So marketplaces do not have the same level of control over every freelancer or agency that's on their platform. In the case of agencies, they can spend on marketing and sales. So their success is even less predicated on how well they fulfill on one platform.
The tools for project management and collaboration are so varied that it's hard to standardise on tools across service providers. The context varies so much between each need that processes cannot be standardised either (or the complexity of standardising towards 'n' possible delivery approaches is so high that no marketplace has attempted it, beyond the garden-variety placeholders they have for process). Also, knowledge workers are different from Uber drivers. Building software or writing content is different from driving a cab around. All these make it difficult for marketplaces to ensure uniform fulfillment experiences.
They do the next best thing - failure guarantees (like escrow and arbitration). There are risk mitigations but no quantum leap in successful outcomes.
As an insider, I've come to believe that data and structural control mechanisms of the present do not allow for a fully automated marketplace that delivers exceptionally good customer experience. The best experience still demands human intervention and it will be so for a long time to come.
Until technology catches up to understand contexts very well (I am bullish that AI will solve this problem within the next five to 10 years), remember this - you still need a wise counsel on your side.
(image credit: Shutter Stock)
(Disclaimer: The views and opinions expressed in this article are those of the author and do not necessarily reflect the views of YourStory.)