CrowdFlower raises $12.5M to deliver better data for better models

Crowdsourcing startup CrowdFlower has raised a $12.5 million Series C round of venture capital, thanks in part to the rise of data science as an accepted practice within many companies. Canvas Venture Fund led the round, which also included existing investors Bessemer Venture Partners and Trinity Ventures. CrowdFlower has raised $28 million since launching in 2010.

Lukas Biewald (pictured above), the company’s founder and CEO, said the last year has “by far” been its best yet in terms of user growth, and he chalks some that up to the fact that data science teams are now popping in all sorts of companies. If these teams are going to produce the best possible results, they need the best possible data to feed their models. Unilever uses CrowdFlower to extract sentiment, location, sex and other information from tweets; eBay uses it to clean up its product taxonomies; and some financial services firms use it to perform market research on investment targets, Biewald said.

“If you have a lot of data and it’s clean data, you’re going to have a really good model,” he explained. Or, to use an old cliche: garbage in, garbage out. Even as advances in artificial intelligence improve machines’ abilities to understand language and images, humans are still a lot better in most situations.

Credit: CrowdFlower
Credit: CrowdFlower

Aside from helping generate more and cleaner data, Biewald said the speed of CrowdFlower is also a big selling point. Historically, companies wanting this type of data or these types of services might hire an outsourcing firm that could take days or weeks. Data scientists, in a quest for quality control, have been known to spend a lot of their own time cleaning up data. Biewald said CrowdFlower solves both of these problems because users can easily specify, via a web interface or API, what they need and when, and CrowdFlower’s algorithms take care of finding the workers and managing the process.

The new funding, he added, will go toward helping make the platform even more self-service to cater to the data science community CrowdFlower is now targeting — a community that by and large would rather get to work without having to interact with a salesperson. It can be a tricky proposition to automate more of the process and still keep quality high, but Biewald there’s still a lot that can be done.

If you’re reading this and think that you don’t remember all this talk of data science when CrowdFlower launched (check out the 2010 video interview with Biewald below), that’s because it wasn’t there. But in the past couple years, Biewald explained, the company came to realize — when they started calling themselves it — that many of its users were data scientists.

That was a welcome revelation, he added, “It gives us a clear understanding of who our audience is.”