In my previous post about Lionbridge, I wrote about how its enterprise crowdsourcing division is challenging the traditional outsourced services model with “business process crowdsourcing” for the enterprise. This managed crowdsourcing strategy adds governance and high quality to the crowdsourcing approach to provide an alternative that Lionbridge says is in the range of 30 percent cheaper than what traditional business process outsourcers charge. So how does Lionbridge do it?
According to Dori Albert, Lionbridge’s enterprise crowdsourcing practice manager, it starts with attracting qualified workers for the company’s private crowd. She explained that these workers are thoroughly screened and tested before they’re accepted into the crowd, and that they’re paid an “equitable wage.” I asked Albert if she could quantify what Lionbridge considers to be an equitable wage, and she said it depends on the task:
I asked her if she could give me an estimate of the hourly wage for workers in the United States, and she said that, too, depends:
If you’re doing low-end data entry, you’re probably talking about federal minimum wage-type work. If you go higher-end, like some search relevance work, you can go up to $15.00 an hour. Testers and translators are typically a lot more than that. We have CPAs doing work, and they make CPA wages in whatever country they’re located.
Lionbridge’s private crowd consists of around 150,000 workers in over 100 countries, so you can well imagine that paying them would require a pretty robust payment system. Albert explained the payment mechanism that Lionbridge uses:
There’s an art to figuring out the appropriate payment for people, and how you actually pay them is very important. We had to figure out how we were going to pay all of these people in currency, all over the world. We pay in four currencies using an international bank that’s located in Ireland—we pay direct to their bank account. It starts with how we sign people up—we have a fairly sophisticated technology platform that’s based on SAP. From the beginning, when people are applying for jobs with us, we take a lot of information from them, including bank details. And they go through a whole qualification process before we even let them into our crowd. That’s another way that we know that they’re more dedicated. We do bank wires direct to their bank account—it’s a combination of a lot of processes and technology that we’ve put in place over the past several years. It’s a sophisticated process—you have to know a lot about the labor laws and banking worldwide.
I asked Albert if she could give me a sense of the geographic disbursement of Lionbridge’s private crowd, and she did better than that—she gave me the percentage figures:
Right now, the highest percentage of our workers are in Europe and North America, and Asia after that. We’re seeing a lot of growth in the Asian countries, particularly in India, because of the demand for English skills. I think those percentages are indicative of the Internet connectivity. Currently, Europe is the biggest, at 42 percent; North America is next at 31 percent; Asia is about 13 percent; it’s spread across a bunch of other countries after that. Europe is the highest because a lot of the work that we do requires multilingual workers, and most of our crowd in Europe is bilingual.
On the topic of privacy and security issues that surround client data, Lionbridge makes a big deal of the fact that it requires all of its workers to sign NDAs. I told Albert I would argue that NDAs are difficult to enforce under the best of circumstances, but under these circumstances, enforcement for a large percentage of the workers would be so close to impossible and could render these NDAs entirely meaningless. I asked her what I was missing, and she responded that she would never consider an NDA to be meaningless:
It’s a binding document. I wouldn’t argue with you about how they would hold up in court, because I’m not a lawyer, especially given that we operate in so many countries. I will tell you that the way we design our work and our tasks is the key—we don’t expose client data that would put our clients in any sort of compromising position. We extract all the PII [personally identifiable information], or we scramble it in a way that the workers could never glean anything from it.
Finally, I asked Albert if it was fair to say that there are certain types of IT projects that Lionbridge’s managed crowdsourcing model would never be suitable for, and she said it absolutely is fair to say that. She noted that there are crowdsourcing companies that do a lot of development work, citing the fact that TopCoder was recently acquired by Appirio. Albert said Lionbridge has chosen to stay away from development:
What we’re finding is that you can componentize development, but it’s very hard to get teams working together. What you have to do to make crowdsourcing work is to not have a lot of collaboration between people. So when you have a development project that requires people to be working together, to be talking to the business analysis team, discussing requirements, that’s not what this model is. Plus there are certain things that companies want locked down in a center, like PII. So you don’t want to work with information where there’s no way to do the work without having the PII in front of you. Those things are best for a traditional outsourcer, because they’re going to be locked down in a center.