Google for Jobs: Disrupting The Recruiting Market?
One of the most challenging problems in business is matching the right person to the right job, and this goes on all the time.
The Bureau of Labor Statistics show that 20-24% of Americans change jobs every year (ADP global research says it’s 27%), which means more than 41 million people are searching for jobs and being recruited into jobs every single year (in the US alone). Several billions of dollars are spent on job advertising (Indeed, Glassdoor, CareerBuilder, and others compete for this market) and even after people apply, companies on average spend approximately $4,000 per candidate on interviewing, scheduling, and assessment to decide if someone is right for a job. We estimate that the entire recruitment market is over $400 billion worldwide, and nearly every employer is a participant.
For job seekers, the search process can be agonizing, difficult, and frightening. Unlike other searches on the internet, a job search is a very personal thing. You are looking for a position that fits your needs, a job with a company that fits your personality and lifestyle, and an employer that is physically close enough that you can commute or relocate without impacting your family and daily life. All these search “criteria” are important, and almost none of this information is embedded in the job description.
While companies like Glassdoor, Indeed, LinkedIn, and others try to give companies branded pages to promote their company culture and wonderful workplaces, most job descriptions are limited, out of date, and often poorly written – making search engines a problematic way of finding a good job. Google has been studying this problem and found that for any given job (e.g., Marketing Manager) there are hundreds of different job titles (“Marketing Manager,” “Digital Marketing Manager,” “Marketing Specialist Level 2 ,” and on and on), often making searches inaccurate or misleading.
For technical jobs (specialized engineers, scientists, manufacturing staff, and even truck drivers) job titles are even more “non-standard.” Google research found that you can search for a simple job like “truck driver” and find that the results obtained in FedEx, UPS, and other delivery services are often different. And even within the same company you find that very similar jobs appear under different titles: FedEx Express calls drivers Couriers, FedEx Office calls them SameDay City Couriers and FedEx Freight calls them City Drivers. They each use different language to describe that very job, so you have to spend a lot of time search each company’s website to find the job you want. Which of course makes it nearly impossible.
Behind all this confusion is the marketplace of applicant tracking systems (the corporate software that companies use to post jobs, manage applicants, schedule interviews, etc.). These products are often old and their search and scoring engines are fairly rudimentary, so all these non-standard jobs on the internet are not only hard for people to search, they are hard to match to candidates. So just as candidates often randomly apply for jobs that aren’t a great fit, the ATS has a very difficult time scoring resumes to decide who to call back.
The bottom line is a lot of headaches and inefficiency in the job market: the average open position receives more than 150 resumes, more than 45% of candidates never hear anything back from the employer, 83% of candidates rate their job search experience poor, and employers still tell us anecdotally that 20-25% of their candidates don’t turn out to be a good long term fit.
If you look at the data collected by the Talent Board on the candidate experience, you find that more than 1/3 of all job seekers spend 2 or more hours researching a single job, it often then takes them an hour to complete the job application, and more than half rate the search process poor or mediocre. We are creating a lot of pain, effort, and complexity for job seekers everywhere.
I won’t give you the history of the Google Cloud Jobs API team (it’s an interesting history), but some very experienced people at Google have been looking at this market for a few years. They realized that the entire taxonomy of jobs is a mess, so one of the things they have been doing is building a “job family taxonomy” that aggregates similar job titles into families of jobs to build a truly useful, searchable, “universe” of jobs, organized by discipline and functional domain. This alone is an enormous undertaking, and one that will benefit our economy and entire society if it really works well.
As the team has been building this out, they started using the taxonomy and search algorithms (it also uses Google’s Natural Language Processing algorithms to read job descriptions and find out that “truck driver” and “delivery manager” are the same job) across millions of jobs to train the algorithm over time. Their experimental work with early customers has been fairly amazing: one client found that searches for “genetic engineering research” jobs barely surfaced a single job before using the Google technology; after using the Google search the perfect job popped up in the first page. And Johnson & Johnson has seen an 18% lift in applies per search on its career site since integrating.
Fig 1: Google Jobs Taxonomy
At the heart of Cloud Jobs API are data models that encode knowledge about occupations, skills, education, employment type, location, seniority, and other career signals. This is Google’s visual representation of the title data model which includes 250,000 normalized job titles.
Remember that Google is quite expert at understanding your search terms, keeping track of all the searches done in the past to get smarter, and then using smart NLP algorithms to decompose your search into a meaningful query and then pull back the best possible result. This is what Google has done for its entire life, so applying it to job search is a natural thing.
The Potential Impact of Google for Jobs
Finding the right employee is one of the most important tasks in business (one might argue it is the most important thing managers do). Yet 61% of top executives surveyed in the Deloitte Human Capital Trends survey told us their companies do not do it well. So if Google can make job sites better, make the candidate search process more accurate, and give us all better data about what good jobs are available in our location, many problems can be solved.
And this problem is getting worse. Our research with Burning-Glass Technologies shows that the jobs of today are shifting away from static “jobs” to “roles,” with much more of a hybrid nature. These new “hybrid jobs” (Ie. IOT engineer, digital marketing experience manager, etc.) do not lend themselves to static job descriptions and simple job titles. They are jobs that require technical, industry, managerial, and integrated thinking skills; they often require skills in communication, persuasion, and teamwork. So more than 2/3 of companies now use job-based assessments (particular games or tests which simulate the job itself) and culture assessments to find the best candidates.
Just as Google continues to improve its search capability for restaurants to be “smarter,” using data like location, similar searches you’ve performed, and demographic data about you, so can Google do the same for jobs. One of the features in the Google for Jobs project is to search by “commute time.” Rather than look for a job within a certain number of miles from your home, Google can use Google maps data to tell you precisely how long it will take you to get to this job.
Imagine, for example, searching for “second line manager jobs in my profession which need my type of experience that are within a 30 minute commute of my home.” This is the type of search that will become possible on your career site with Google’s technology.
Just as Google has continued to improve its technology for targeted advertising, so Google for Jobs could do the same for job posting and search. Imagine an employer that buys a job ad that targets professionals with a certain level of experience in a certain city who have worked for a certain company. While Google may not have all that data in your profile, much of it is discoverable on the internet so one could imagine Google’s search engine (and ads) to get very smart over time.
A Potential Benefit to the Economy
Imagine if people could find jobs faster with a better fit. Imagine if employers got fewer, more appropriate candidates. Imagine if data about what jobs are “most in demand” was easy to find. The program Google has undertaken has the potential to drive economic growth, personal career growth, and business productivity for almost everyone. And it’s available today, simplifying machine learning by making it universally accessible to the Talent Industry.
The problem is still daunting, and I know Google still has work to do. But the team working on this project is very passionate and experienced, and already their work is starting to show off. I, for one, will watch the Google for Jobs project with great anticipation. The potential could be enormous.
Example of Improved Job Search Accuracy with Google for Jobs
Fig 2: Visual Representation of Google’s Job Taxonomy
Description: This is a side by side comparison of a query for “Accounting Manager” on Johnson & Johnson’s career site. On the left, you see results from the query running through the Cloud Jobs API, which correctly returns relevant Finance jobs to the job seeker. On the right, you see the results for the same query without the Cloud Jobs API – note how Sales jobs are returned due to a misinterpretation of the query.
As used in this document, “Deloitte” means Deloitte Consulting LLP, a subsidiary of Deloitte LLP. Please see www.deloitte.com/us/about for a detailed description of our legal structure. Certain services may not be available to attest clients under the rules and regulations of public accounting.