ZipRecruiter raised $156 million in Series B venture funding with plans to expand its machine learning and artificial intelligence infrastructure and grow internationally.
The online recruiting marketplace, which has raised $219 million in total venture funding, hatched in 2010 as a service that enables companies to post job listing in multiple places. Since then, ZipRecruiter has pivoted from a volume play to one focused on using machine learning, deep learning and artificial intelligence to bring quality candidates to companies.
ZipRecruiter has received more than 430 million applications for jobs at more than 1.5 million businesses since 2010. The company has a large small business customer base, the its enterprise footprint is growing at a rapid clip.
We caught up to Ian Siegel, co-founder and CEO of ZipRecruiter to talk about his company's AI plans and landscape.
Where does ZipRecruiter sit in the HR technology space? Initially, ZipRecruiter was a way to enable companies to post one job listing in multiple places, but generally the space was dominated by companies like LinkedIn and Monster. "Today, there's a new crop for job listings like Glassdoor, Facebook, Google for Jobs and Indeed. There's access to an unlimited population for candidates," said Siegel. "For a long time we were in the volume business and it worked well, but then there was an inflection point where you had too many candidates. Once you get 150 candidates you go 'I wish I could just get 10 really qualified candidates.'"
How does AI help? With a pivot to quality over volume, ZipRecruiter opened shop in Israel and now has 50 engineers focused on AI and machine learning matching techniques. "Now we are surgical and can cherry pick matches," said Siegel.
But you need a large data set. Siegel noted that all the volume from ZipRecruiter's inception gave the company a large corpus of data. "We have all the job descriptions and resumes. We have the interactions between candidates that apply and come through our systems. We know things like did the employer email, call or look at a resume more than once and whether it was forwarded," said Siegel. "With these large populations interacting we are able to do lookalike analysis on candidates."
Deep learning with 64 dimensions of data. Siegel said that data corpus has allowed ZipRecruiter to develop algorithms that can build matching techniques based on 64 dimensions of information.
In other words, ZipRecruiter's initial focus on volume may not have been all that sophisticated, but it ultimately enabled its data science efforts.