Interview with Steve Goodman (Bright)
Bright is a job searching tool for both employers and prospective employees.
I interviewed Steve Goodman, Bright CEO to find out more. This is the hundred and sixty seventh in a series of DW startup interviews. Big thank you to Steve!
How would you describe Bright in under 50 words?
Bright unlocks the power of artificial intelligence, machine learning, and big data to establish a one-to-one signal between job seekers and successful placements.
What gap in the market did you discover that persuaded you to launch Bright early 2011?
We noticed that the Internet had democratized the job application process to the point where employers now receive 6x more resumes than just 10 years ago. Companies like Coca Cola now receive over three million resumes a year, while just a few years ago it was less than 500,000.
The Internet has made it almost too easy to apply for a job. How do employers quickly surface the best candidates from a pile of three million resumes? Bright set out to solve this problem. Bright provides what’s akin to a GPS for job seekers – guiding them to apply for the jobs where they are qualified.
You employ over 30 people. Ironically, have you had any difficulties in finding talented developers and managers?
As of March 2013, we employ about 35 people (and we’re hiring). We haven’t had a hard time attracting great talent. Bright.com helps people find jobs. We are making a substantial impact on people’s lives. We aren’t here to simply sell widgets or make money, we’re on a mission to move the labor markets faster.
We don’t pay top dollar for talent; we are interested in working with people that want to make a difference in the world. If you are singularly focused on earning top dollar, there are plenty of opportunities in Silicon Valley. If you want to impact the world by helping people find work for themselves and their families, we’re a pretty good spot. This value prop attracts the talent we need to forward our mission. Our interviews are as much about culture fit as nailing a few lines of Python.
How long did it take to put together the initial version of Bright?
Over a year and a half. It took about 15 data scientists and engineers, over 100 talent recruiters across industries, and the largest scientific resume to job description study in history. Inputs included over 2 M resumes, 8 M job seekers, and more than 20 M job descriptions.
How much of your development time has gone into the machine learning algorithm ‘Bright Score’?
Most of it. The Bright Score algorithm looks at hundreds of features in a resume (i.e., gaps in resume, time in job, working at a competitor to the company the job seeker is apply to) against a job description. We’ve logged almost 75,000 man hours into development of the algorithm. We’re altruistic with its development and it learns on its own over time. There’s very little manual scaffolding supporting it. The more resumes and job descriptions it sees, the stronger it gets. We’ve now processed more than one billion Bright Scores.
What technologies have you used to develop Bright and which were the most technically challenging development problems?
Lots of challenges. We are building something that hasn’t been developed before. The biggest challenge is reducing an N-squared matching problem to a manageable dilemma. The Bright Score matches N candidates against M potential jobs. We need to apply appropriate intelligence to make this tractable.
Most of our work is steeped in statistics and inferences, but processing unstructured text in a timely fashion that can provide an accurate Bright Score in moments is hard to do. We use standard development / storage technologies, but building the machine learning algorithm is the root of our challenges. On the data science side we use SciPy, NLTKs, MATLAB, Lucene, and Hadoop’s MapReduce. LAMP stack for the website and then CouchDB, and non sql database Neo4J.
Do you feel your company is more useful to the employers by helping them short-list applicants or for prospective employees?
That’s like asking Match.com if they do better for men or for women. Our goal is to create an efficient marketplace, and there should be meaningful balanced value on both sides.
You’ve received $6M in venture capital, how long did it take to secure the funding?
Not long. Our team has a pretty good track record of raising money and building good businesses in Silicon Valley. We’ve worked with top tier VCs and angel investors in the past.
Do you have any features in the pipeline to make job searching even easier and more efficient than you have already made it?
Yes, of course. But I’d have to kill you (and I’d be killed by our PR team) if I told you what they are.
Has Bright got the feedback and growth you expected since launching in February 2011?
Far exceeded. From a standing start of zero visitors less than two years ago, in March ’13 over 4 million unique job seekers will visit Bright.com. 20,000 job seekers a day register with Bright. And we’re almost profitable while sustaining our strong growth. Our business outlook is strong.
How do you differentiate yourselves from your competitors?
The Bright Score and our people. Over the years many companies have developed keyword matching systems. For example, LinkedIn uses a nice keyword matching system. The problem is that when a talent recruiter searches on “Java”, returned results will include people who took a class in Java, people with between zero and 20 years experience in Java, and even Barista’s that work at Starbucks!
Boolean search tools used by talent recruiters are really poor funneling systems. The Bright Score fundamentally solves this problem because it looks at the whole resume against the job description. It saves time and money for talent recruiters. The other major differentiation is our team: not just the pedigree and academic backgrounds, but the passion and effort they put into the business. Bright is woven into the fabric of our lives.
When watching the news do you find yourself relating stories back to how they will affect your business?
Every day. What are the top stories in the news these days? Jobs, the economy, employment trends etc. Our business could not be more on topic to the state of the economy or current events. Just check out Bright Labs to see how.
What is your primary focus in terms of new developments at the moment?
Similar to Google, our algorithm is the core of our business. We focus like a laser on it. Google’s algorithm is substantially stronger today than in 1998 when they first started. And Google continues to invest most of their R&D dollars into it. We do the same thing. Our Bright Score is only going to continue to get better.
Has the high unemployment rate increased interest in your company due to more people looking for jobs, or hindered the company due to less companies looking for new employees?
Neither. We are a technology innovation in the labor market regardless of the state of the economy and employment levels.
Where do you see Bright in 5 years time?
We are a core algorithm / technology company first and foremost. Ultimately, the Bright Score will be part of every job application evaluation process by talent recruiters. Similar to a FICO credit score in the credit industry. Obviously job seekers get a different Bright Score for every job, but the comparison to the FICO score is that it’s an industry standard. Before a bank reviews a credit application they check the applicant’s FICO credit score. The same should happen with employers and the Bright Score. Why waste time on a job seeker’s full application if they aren’t at least minimally qualified?
With the knowledge you have today, would you do anything different if you were to start-up Bright again?
The Bright Score coupled with the social graph is where job placement results are accelerated. But we spent a bunch of money and time on the hypothesis that “social recruiting” by itself was a big breakthrough in HR. We realized that it wasn’t true.
The social graph is important for sure, but just because you’re friends with someone on Facebook doesn’t mean you’ll recommend them for a job. In fact, are most of your Facebook friends really in that small circle of trusted friends that you would even tell you’re looking for a job? We’re huge believers in the social graph, but it’s not the only, or even best, factor in identifying whether someone is qualified for a job. It needs to be coupled with an algorithm like the Bright Score.
What is the biggest hurdle you have faced or are still facing?
Building a startup is like hitting from an erratic automatic pitching machine. You have two choices: swing or duck. Keeping everyone in the batter’s box, focused on the ball, and swinging at the best pitch is the constant struggle for us and all early stage companies.
What advice would you offer to any soon to be startup founders out there?
Try something, fail fast, try again. If it’s not working try something new. Fast. Think 20-30% of the time, execute 70-80% of the time. That is, stop thinking, and start doing. Revenue is the best deodorant. Get customer feedback early and often…Read Steve Blank’s product development stuff (Lean LaunchPad)…Hire people smarter than you in their areas of expertise. Surround yourself with smart capable leaders and get out of their way…Hire someone to call bullshit on your bullshit. Work your ass off, don’t take yourself too seriously, and make sure you get to your kid’s baseball game.
What are you most excited about at the moment?
My kid’s baseball game at 5:30 tonight.
Can you convince the reader to start using Bright in under 50 words?
Finished reading? Check out Bright!