At this year’s annual Google I/O event, CEO Sundar Pichai closed his keynote address by unveiling Google for Jobs, an innovation designed to make it easier for people to find a job by returning more relevant search results.
Like most things involving Google, it has been surrounded by hype, with Sundar dramatically announcing that the project “transcends jobs” and hopes to empower economic prosperity for the world. The audience roared its approval, but I have to admit I was underwhelmed by the demo. I’ve come to the conclusion that Google for Jobs is more of an incremental (yet necessary) step forward as opposed to an exponential game-changer.
The companywide strategy that Sundar shared involves shifting to an “AI-first” approach, using artificial intelligence to improve function across all of Google’s platforms. You’ll begin to see applications that do things for you automatically – Google Photos will identify who is in your pictures and categorize them, Google Lens will identify objects (flowers, restaurants, etc.) using your camera, and Google Home will recognize your voice and customize your experience based on your preferences. I had thought Google for Jobs was going to automate the manual process of searching for jobs, but that isn’t the case.
What It Does
The reason why search results are so random on most job portals is that they haven't done the hard work of teaching their software the difference between, say, an Administrative Assistant and an Office Manager. While these two jobs might share a lot of the same keywords, the level and content of the role could be significantly different.
Check out this simple search on a major aggregator for an Administrative Assistant. I would argue that only 1 out of the top 4 results are relevant:
In order to return better search results, a product has to analyze the full job listing and have a deep learning algorithm that allows it to properly classify the job. This is accomplished by bouncing the information off huge sets of data to compare keywords, alternative job titles, job levels, salaries, competencies, and other natural language phrases that, when knitted together, magically return a more relevant job search result.
That's what Google for Jobs is doing. And it’s doing it with the same infrastructure that enables AI, Machine Learning (ML) and advanced search capabilities within the other products in their ecosystem.
A Step in the Right Direction
The biggest problem with searching for a job today is not the act of typing in search terms or filtering by location. Most job sites do a decent job of that today. It’s knowing when the perfect job is available – even when you’re not looking.
I'm referring to that platinum "open candidate" we all talk about in the recruiting industry. That person who is a high performer who could be open to a new opportunity if it's presented at the right time and in the right way. According to a recent Linkedin study, these passive candidates make up 90% of the job market!
But instead of catering to this crowd, Google for Jobs is simply optimizing for the task of manually conducting a job search. Past Linkedin research estimates that this addresses only about 12% of the total job market.
In an AI-first world, Google should be gunning for a better matching algorithm – something that alerts and informs people about jobs automatically so they don't have to conduct a job search at all. This would help both the passive and the active candidates by saving them time and curating roles that match their interests.
To be fair, I think Google has a bigger vision than simply improving search results. But like most big data projects, you can't tackle the big vision until you've properly mapped all the data. That's what I hope Google for Jobs is – a project to map the complex and highly variable world of job titles and job-related information for the benefit of some future application of AI.
Google Jobs API
Probably the most encouraging announcement made was that Google has opened their technology up to others via the Google Jobs API. They've partnered with the likes of LinkedIn, Career Builder, and Glassdoor, and are allowing staffing agencies and direct hire organizations to use the technology as well. Together, these entities will help build out the data set needed to feed the mapping project.
With Google making their technology somewhat open source (notice I didn’t mention Indeed as a partner), a recruiting leader now has the ability to put the power of Google search into their own career site via the Google Jobs API. According to Sundar, this concept has already delivered to Johnson & Johnson an 18% lift in applicant conversion rates on their corporate career site. Assuming no drop in quality, those are encouraging numbers!
By partnering with these organizations, I believe Google will not only deliver an incrementally better search experience, they'll also be building the foundation to do something much more innovative from an AI perspective.
What Would Exponential Improvement Look Like?
I'll continue to monitor Google for Jobs and even evaluate the Google Jobs API for clients where it makes sense, but what I really want to see is that AI layer that will make this data exponentially more valuable for the global labor market.
To illustrate what an exponential improvement in job search might look like, below is a scenario that leverages some of Google's core products while delivering an amazing and proactive job notification experience. (This scenario would make use of the Google Home product or the Google Assistant app.)
Me: Hey, Google, what's my news for today?
Google: Good morning, James. Before I deliver your news, would you like to hear about a new job opportunity that just surfaced? It looks like it would be a promotion for you with a 30% increase in compensation, and it's located just 10 minutes from your house.
Google: The name of the company is Cielo. They have been a top RPO firm for several years running, and they're growing at 20% year over year. They're looking for a Technology Enablement Director to expand their technology vision to their global markets in South America and Asia. Would you like to learn more about this role?
Me: Yes, it sounds interesting.
Google: OK, would you like me to connect you to a recruiter by phone right now or should we schedule something for later?
Me: Later. Can you check my schedule and find a free 15-minute slot sometime after lunch today?
Google: OK, I checked your calendar. It looks like you and the Cielo recruiter have an opening at 3 p.m. today. I'll send an appointment request along with the details of the job. Please check your calendar and confirm that you're available by accepting the invitation.
Me: Excellent, will do. Thank you!
Google: You're welcome. Good luck with your conversation with Cielo today! Would you like to hear your news now?
AI is Now, Not the Future
While this scenario might sound a bit futuristic, I actually don't think we're too far off from it being reality. Especially if Google is able to harness the jobs data and combine it with their existing AI and voice capabilities.
Despite the announcement of Google for Jobs falling a bit flat for me, I have an incredible amount of faith and optimism that AI will transform recruiting and job search globally.
For now, we’ll start to see job listings at the top of the page when conducting a job search on Google. This is a small improvement, but for me, the innovation has to be in the matching and delivery of jobs to people, not in the task of manually conducting a job search. And the way to that lies in AI.