How We Found the Best Job Sites
A 2015 Pew Research Center report revealed that 79 percent of Americans seeking employment in the previous two years used online resources as part of their job search — more than personal connections or professional contacts, and twice as much as hiring agencies, traditional ads, or job fairs. Nearly one-third of job hunters said the internet was their most important resource. Convinced that finding the best job site is important? We are.
Our goal was to evaluate general-appeal job sites; nothing niche or industry-specific (although we touch on a few below). We had a few priorities.
- We wanted to be able to filter by date. Job sites try to list the most relevant search results first, no matter when they were posted — but we at least wanted the option to sort for fresh posts so we weren’t sifting through the same jobs again and again. Even better: Sites that also let us view only the new posts since our last visit.
- Mobile apps were a major plus. That same Pew study found that of the people who owned a smartphone, 94 percent used it to research job postings, and 50 percent used it to fill out an online job application. Half of those job seekers also reported encountering “challenges navigating the job search on a mobile device” because content wasn’t displaying properly; another third “had trouble entering a large amount of text needed for a job application or had difficulty submitting the files or other supporting documents needed to apply for a job.” Mobile-friendly sites can fix the first problem. Only an app can conquer both.
- And we required the option for daily email alerts. Alerts deliver relevant opportunities straight to your inbox — no searching required. It makes the job of looking for a job one step easier.
First up: We tested pure usability.
Hunting for jobs requires a lot of time staring at a screen, mining job boards for promising postings. Layout inconsistencies from post to post, button overload, and manic pop-ups are particularly frustrating. Worse is confusing messaging. In one search, we were repeatedly asked to refine our location from Los Angeles, California, to “City Terrace Branch County of Los Angeles Public Library.” Huh?
In another, we clicked “Apply” and were congratulated on our successful application — even though we hadn’t uploaded a resume or filled out a single field.
We preferred functional design elements — like “save” and “email” buttons — that were intuitively placed and consistent throughout the site. A clear path to apply, without numerous clicks or re-directs, was also good. Bonus points if a site managed to avoid cheesy stock photography.
Only the best made it through this most basic test. The rest, we dropped from the running.
We tested each remaining site’s search algorithms.
Tejaswi Tenneti, a software engineer at Apple who previously worked for Monster, gave us the low-down on the unique nature of job site search engines compared to, say, Google. “Google can mix and match different queries and is built to cast a wide net,” he explained. “It’s not always clear what you’re searching for, so Google just compiles everything into one place. But with a job search engine, the site knows people are looking for jobs and the algorithm is much more constrained — and users have higher expectations.”
Our expectations were no different than any job seeker’s: the highest number of the most-relevant job posts displayed on the first page of results, with minimal filtering.
“A good search engine should be able to guess what the user is looking for, even with limited information,” Tenneti said. That’s because job site developers are responsible for building “concepts” or “grids” that parse through a job post for key information like education requirements, experience, skills, etc. When a job seeker enters in a query, even a broad one, the site links it to a concept, and serves up results that match.
If the concept isn’t good, the results will be inaccurate — for example, you shouldn’t see a job posting for a truck driver after searching for “software developer” because the education and skills in those concepts aren’t the same. More elaborate, stronger concepts have more information attached to them, and therefore return more nuanced results.
At first we thought search terms and location would matter (maybe one job site would have better results for algebra teachers in rural Kansas than another), but we quickly realized that it was more important to evaluate how well the algorithms worked compared to each other, regardless of the job we wanted or where we lived. So we devised a series of tests to compare each site’s algorithm from different angles: freshness, frequency, overlap, and quality.
Overall freshness came first.
According to a 2014 study conducted by SmartRecruiters, the best day to check job sites for new listings is Tuesday. That’s when companies and hiring agencies post the most. Our findings echoed this: Tuesday was the most popular day for new postings, followed by Friday. Sunday was the slowest day.
We searched the seven remaining sites for “Nurse Practitioner,” “Financial Advisor,” and “Software Engineer” in Dallas, Texas — jobs in three of the fastest-growing industries in one of the nation’s best job markets. We recorded the total number of postings for one day, as well as the oldest and newest postings within the first 50 results. (According to Tenneti: “Most users don’t go past 40 or 50 jobs.”)
We let these first impressions guide us, and quickly ditched both CareerJet and LinkUp. Both had multiple dead links and extremely outdated posts — one job post from 2014 was still kicking around the first 50 results of LinkUp. That left us with five final contenders: Glassdoor, JobisJob, LinkedIn, CareerBuilder, and Indeed.
We looked at the frequency of new posts.
Every day for two weeks, we tracked the number of new search results posted for Nurse Practitioner roles. For this test, we were interested in the numbers only: Which sites had the most frequent new job postings?
Glassdoor was the winner here, with an average of 20 new posts a day. The closest contender was LinkedIn with 10, and the worst was CareerBuilder at a measly five.
Then we mapped the overlap from site to site.
“There are two models of job sites,” Tenneti told us. “There are some where people can post openings directly to the site, and there are aggregators that pull from other search engines and company websites. Most job sites try to be both.”
All five of the job sites we were testing could both host job posts, as well as aggregate from other places on the internet — including each other. In a perfect world, this means you could go to any one site and they’d all have the same postings.
We do not live in a perfect world.
To find which had the best coverage, we tracked postings from each job site to see where else they popped up. Indeed won this round: It included 65 percent of the postings listed on other sites. Glassdoor followed close behind with 60 percent overlap. CareerBuilder was by far the worst, with only 20 percent.
Of course site popularity influences the number of postings it receives organically. The more companies post directly to a site, the less it needs to rely on scraping. Still, even lesser-known sites — like JobisJob, which had 45 percent overlap — can compete if they have a superb algorithm.
And assessed email alerts.
We received email alerts for Nurse Practitioner roles in Dallas, Texas, for two weeks, and our guidelines were strict. Without additional filtering, the alert had to match the location, contain the words “Nurse Practitioner” in the title, and actually be related — no spam.
At first, JobisJob was in the lead, but then multiple postings for Dallas, Iowa (about 700 miles away from Dallas, Texas), found their way into our inbox. Then, Uber tricked JobisJob into featuring its posting titled, “Become an Uber Driver Partner — Instead of a Nurse Practitioner.” No good.
JobisJob wasn’t the only site to experience problems. We tried signing up for daily alerts with CareerBuilder twice — we even received a confirmation message — but no alerts showed up in our inbox until 11 days into testing. The alerts we did receive featured the same three jobs four days in a row and didn’t match the title or location we specified. CareerBuilder was already on its last leg, so when it performed the worst in our last test, we officially bid it adieu.
Lastly, we cracked down on quality.
This was our most important test: drilling down into each job posting to see how accurate and useful it was to our search query. What good is a user-friendly, high-performing site if all the postings are for spam jobs or they don’t link to the original job posting?
We spent a day assessing the first 10 postings (not including sponsored or advertised posts) on each site. We looked for five factors:
- The location — Dallas, Texas — was accurate.
- The words “Nurse Practitioner” were in the title.
- The post provided a direct link to the original job posting.
- The date on the job site matched the date of the original posting.
- The application stage was less than three clicks.
Glassdoor and Indeed were once again in the top spots, able to hit 40 percent of our criteria. Yep, that’s right: 40 percent. Even those of us who struggled with statistics in high school know a grade like that can land you in summer school.
So what gives?
Stay tuned for ZipRecruiter.
We’re continuing our testing, this time including ZipRecruiter. All signs point to it being a top contender, but we’ll have official results by February, 2017.
Turns out, it’s incredibly difficult to track down original postings. Some sites linked to a general company page — not super helpful, especially if a large company has hundreds of job openings to sort through. But most just linked to other aggregators, like Resume Library and ZipRecruiter, which in turn might link off somewhere else.
And that original posting is important. Aggregators that link to each other create a game of telephone: The more information that passes through multiple parties, the higher the probability for errors. Having access to the original post saves the time and frustration of fact checking if the job is even still open, let alone having to navigate yet another job site.
Come on, job sites. You have one job! And even the best can’t seem to do it very well.
But then we spoke with Steve Dalton.
He’s a program director for Daytime Career Services at Duke University’s Fuqua School of Business, and the author of The 2-Hour Job Search, and he confirmed our worst suspicions: “It’s the black hole everybody thinks it is. For every one person hired through an online job application program, 12 are hired by an internal referral, according to a 2012 hiring study at the New York Fed.”
So here we were, hours of testing and data behind us, wondering what job sites are even good for if not helping you land a job.
According to Dalton, job sites are the key to finding meta information.
“The literal information you find in online job postings is not that helpful, but the information that the postings suggest is very helpful,” he said. “They can tell you which companies are currently hiring, which companies are looking for people in the cities you’re interested in, and which companies are more time-sensitive. They’re just terrible at getting you interviews.”
What you do with that information could mean the difference between rejection and acceptance.
“Remember that applying online and networking are two separate systems,” Dalton said. “I was doing a talk at a school and heard from a student who applied online for a position at a consulting firm. After he applied, he started networking, did an informational interview, and found an advocate who referred him over to HR. He eventually got an offer. On the same day he got the offer, he also got an email from the company’s automated job-posting system expressing their regret that they chose not to pursue his candidacy any further.”
And that perspective — that it’s more about the context and tools a job site can give you — made us much more confident in naming our top picks.