Traditional models of recruiting are a mess. At companies like Google, quite literally millions of people apply for open roles, forcing hundreds of recruiters to sift through thousands of resumes per job opening. Worse, in their race to process through those applicants, recruiters often use fuzzy and subjective impressions of a candidate to match them to a role, rather than clear and unambiguously objective facts.
Uncommon.co, which launched today, is hoping to change this manual model by using artificial intelligence to identify exactly the requirements for job postings and matching those jobs perfectly to qualified applicants. Also today, the company announced that it has raised an $18 million Series A round from Canaan Partners, Spark Capital, and Zeev Ventures.
President Amir Ashkenazi and CEO Teg Grenager founded the company in 2016, after working together on Adap.tv, which sold to AOL for $405 million in mid-2013 and was the largest acquisition AOL had made up to that time (AOL, now Oath following its merger with Yahoo, is TechCrunch’s parent company). Adap.tv was a programmatic video advertising network that allowed ad buyers quick à la carte access to targeted audiences with a real-time bidding pricing system.
Ashkenazi and Grenager started Uncommon.co when they realized that some of the insights they had learned from programmatic advertising as a marketplace could be applied to recruiting. Grenager studied AI at Stanford as a PhD student and has been working the past two years building out the models used for understanding hiring. In fact, the founders told me that they had ingested around 5 million job postings and 50 million resumes in order to tune their models.
Grenager called the new platform the “world’s first merit-based talent marketplace.” Ashkenazi explained that there are “two important transformations that we are making in the recruiting industry.” The first is “Programmatic recruiting based on qualification data, which means that for the first time, companies will be able to subscribe to a stream of only qualified applicants.” The second transformation is that job seekers will have a much better shot at getting the eye of a recruiter if they are actually qualified for a position, rather than merely thrown into a pile of resumes that might never be read.
For companies using Uncommon.co, recruiters will set up a job description which includes traditional factors like skills, industry background, experience level, as well as factors that have been requested from customers such as average length of duration at former employers. That job description then becomes a posting on the web where job seekers can apply for a role.
It’s here that the intelligence starts. Uncommon.co uses artificial intelligence and NLP to evaluate the resume and compare its information to the description requested for the job. Applicants who perfectly match the qualifications are set aside for recruiter review, while other applicants are placed in an “unqualified” pool, ranked by how close they fit with the job requirements. The company has used its own software for its own job openings, and found that for one position they had about 30 qualified applicants from hundreds that applied.
There were three concerns I had about the product. One was whether the lack of specific words on a resume might disqualify applicants. Grenager explained to me that the company has spent months agonizing over how to create robust intelligence models to evaluate resumes.
Using an example of someone who wanted to work in product, he explained that “if you look through those resumes, it hardly says product manager, you could also be a product marketing manager, or a product owner, or a founder, or a product designer. If you look at it holistically someone might have four years of product management” even if the titles might not be directly connected.
The second concern I had was whether recruiters can competently set the right job description. Every recruiter wants more highly qualified candidates than a position really requires, which is why you see job postings for a new programming language demanding “five years of experience.” The founders quickly discovered this issue, and solved it by giving instant feedback to recruiters on how their choices are affecting the size of the pipeline. So if you require an applicant to hold two PhDs, you might find that there are literally only a handful of applicants in the world that are going to be even possibly qualified for a role.
Finally, with diversity issues being top of mind in the Valley today, I asked whether there might be unconscious bias in the algorithm powering the platform. Ashkenazi explained that he has seen “unconscious bias sneaking in when there isn’t enough data.” Grenager said that “Our qualifications are all around three dimensions, skills they bring to the table, job roles they have been in, and the education they have gotten.” By using fact-based, objective data as opposed to recruiter impressions or cultural matching, the idea is that Uncommon.co will improve the hiring funnel for companies while avoiding “black box” hiring algorithms.
When it comes to the business model, Uncommon.co is eschewing traditional listing fees or cost per click models in favor of something it is calling “cost per interested & qualified” applicant or CPIQ. The idea is that companies should only pay for applicants that meet their own criteria, rather than any person on a job listing board who clicks a hiring link.
For these two programmatic ad founders, the hope is that as their platform reaches scale, there will be real-time bidding for certain qualifications that will determine the price of a listing. The company is charging a nominal flat-rate price today as it scales up, and counts companies like Gap, Aflac, and Lyft as clients.
Dan Ciporin led the investment for Canaan, and has known Ashkenazi for almost twenty years since the two worked together in 1999 on Shopping.com, where Ashkenazi was CTO and Ciporin was CEO. “It was just a very special experience because of Amir, and we became very close in the way that people become close in trench warfare,” Ciporin said. He sees a connection running through all three of Ashkenazi’s companies. “It is helping to qualify things that are raw data into much more refined data, and really enabling people to access information that they want rather than having to do that themselves.”
Uncommon.co is entering a crowded space, but with $18 million in the bank, a group of founders who have returned hundreds of millions to shareholders, and an artificial intelligence platform to scale up quickly, the company hopes to revolutionize hiring and make it easier for both sides of the market to get the applicant — or job — they want.
Update: We have switched the titles of Ashkenazi and Grenager. The titles were changed during the announcement of the fundraise.