Start now

What is the best way to find a job?

On this page
  1. What to do
  2. The evidence
  3. Belief checked
  4. Where sources disagree
  5. Common questions

The short answer. The evidence we checked does not show that any single method is the best way to find a job, and it does not test combining them. So our advice is to run several at once: apply to jobs that fit, ask people you know for referrals, and contact employers directly.

Applications people sent on their own were the top source of hires, at 43% to 52%, and referrals held near 18% (Ashby, a vendor of applicant tracking software, data on nearly 250,000 hires from January 2021 to June 2025). On this page, a vendor is a company that sells hiring or job-search products.

Some randomized experiments of coaching and job search assistance found they raised employment. But a randomized trial found that people sent motivational emails found jobs less often. And a meta-evaluation of active labor market policies (public programs to get unemployed people into work) for unemployed people over 50 found a slightly negative average effect on finding a job across those policies. None of these compares one search method against another.

What to do next (our advice)

  1. Write a plan that uses several ways of searching at once. In a randomized study of 1,097 unemployed young people in South Africa, those who made a detailed plan diversified their search and had more job offers (see "What is the first step to a successful job search?" below).
  2. Keep applying to postings you could do, and expect long odds on each one. The offer rate on applications people sent on their own was 2 in 1,000 (Ashby's vendor data, 2024). So silence on a single application says little about you.
  3. For each employer you want, look for someone who can refer you, ideally someone in the same kind of role. Referrals are more likely to lead to interviews when the person referring works in the same job function as the role (Ashby's vendor data on 54 million applications to March 2026; this is not an experiment).
  4. Include former coworkers and acquaintances in your outreach alongside your close contacts, and ask directly. Records from Austria and experiments on LinkedIn support looking beyond close friends (see "Who should I ask?" below). The LinkedIn experiments tested suggested connections, not requests for help. Across a range of direct requests for help, people underestimated by as much as 50% how likely others were to agree (the first three of a series of studies by Flynn and Lake, 2008).
  5. If your search is narrow, widen the kinds of job you consider. Narrow searchers who were shown alternative occupations got twice as many interviews as similar narrow searchers in the control group (the Belot, Kircher and Muller experiment, July 2016 IZA discussion paper; see "How to optimize your job search?" below). What Works Growth's summary of that experiment found no significant rise in jobs found.
  6. If structured job search help is open to you, consider it, and look for help that teaches search skills and also helps you keep going. The US Department of Labor's reemployment program for people claiming unemployment benefits centers on an in-person meeting with trained American Job Center staff (US Department of Labor fact sheet, page read October 2026). So an American Job Center is one place to ask (see "Is structured help worth it?" below).
  7. Contact employers directly as well as applying. Among people aged 45 to 70 who had been unemployed some time in the last five years, 48% of those back in work had contacted employers directly, against 37% of those still looking (an AARP Public Policy Institute report, as Money reported it, page read October 2026). This compares groups and does not show that the contact caused the job (see "Where sources disagree" below).

Gap: our checked sources don't test how to divide your time between applying and reaching out, or how many applications a week is enough.

What the evidence says

Is it still worth applying?

Yes, as one part of a search, but expect long odds on each application. The vendor figures on this page come from firms in the hiring and job search business. They are drawn from platform data, from their users or, for MyPerfectResume, from a survey of US workers, and they are not experiments.

  • Among unemployed US job seekers in 2024, 55.4% sent out resumes or filled out applications, 50.9% contacted an employer directly, 23.2% contacted friends or relatives and 9.2% used a private employment agency (US Bureau of Labor Statistics, Current Population Survey, 2024).
  • Asked which resource mattered most in their most recent search, recent US job seekers most often named online resources (34%), then close personal connections (20%) and professional or work contacts (17%) (Pew Research Center, 2015 survey).
  • The offer rate on applications people sent on their own fell roughly 70%, from 7 in 1,000 to 2 in 1,000 (Ashby's referrals report, from a vendor of applicant tracking software, 38 million applications from January 2021 to December 2024). Ashby says the fall came as the number of those applications tripled.

Gap: our checked sources don't give the share of hires that come from applications or from contacts for mid-career professionals in particular.

Do I need to know someone?

Referred applicants are more often hired, but people who get referrals may already differ from those who don't, and more networking is not shown to be enough.

  • Roughly 40% of referred candidates reached an interview, against 3% of those who applied on their own (Huntr, a vendor, citing Ashby's analysis of more than 38 million applications; page read October 2026). Huntr cautions that referred candidates are a different, already-selected group.
  • Referred applicants were more likely to be hired than nonreferred applicants despite similar skill characteristics (a Quarterly Journal of Economics paper on personnel data from nine large firms in call centers, trucking and high-tech).
  • Networking harder did not predict finding work, or finding it sooner, beyond what people's other searching already predicted (Wanberg, Kanfer and Banas, Journal of Applied Psychology, 2000).
  • An online networking course called BRIO made people more confident in their ability to network, and that confidence led to higher-quality jobs when they found work again (a field experiment with 491 unemployed job seekers, published by Wanberg and colleagues in 2020). In that 2020 BRIO experiment, participants averaged 46.92 years of age and had been unemployed 6 months or less. Those who completed the course and were less outgoing (lower in extraversion) gained the most, both in that confidence and in the quality of the jobs they found.

Who should I ask?

Former coworkers and people you know less well are worth including, not only close friends.

  • Workers who lost their jobs when their firm closed, and who had a direct link to a former coworker, were twice as likely to be hired by the connected firm as workers without a link (a study of former coworkers in Austrian administrative records of private-sector employment, page read October 2026).
  • The researchers report causal evidence that weak ties (people you know less well) help people move jobs (randomized experiments on LinkedIn's People You May Know feature covering over 20 million people over 5 years, described by the Stanford Digital Economy Lab, page read October 2026). In those LinkedIn experiments, moderately weak ties by mutual connections, and the weakest ties by how much people interact, created the most job moves.

Is structured help worth it?

Some tested programs helped people out of work and some did not, so the kind of help matters.

  • Across 47 job search programs evaluated by experiment or quasi-experiment, the odds of finding employment were 2.67 times higher for participants than for control groups (Liu, Huang and Wang, a 2014 meta-analysis in Psychological Bulletin; a meta-analysis is a study combining many studies). In that meta-analysis, job search programs promoted employment only when they included both skill development and motivation enhancement. The programs also worked better for younger and older job seekers than for the middle-aged, and for the short-term than the long-term unemployed.
  • Motivational emails caused a moderate but significant reduction in search effort and job finding (a randomized trial of 200,720 newly unemployed Swedish job seekers, 2026 IZA discussion paper). Emails alone are not a program that also builds skills, so this does not contradict the 2014 meta-analysis. But it does show that encouraging emails on their own can backfire.
  • Access to one-on-one intensive services (assessments, coaching and career counseling) raised earnings by between 7% and 20% over three years (a randomized trial in 28 randomly selected US sites, described on Mathematica's site, page read October 2026). An NBER study of an experimental Nevada job search assistance program, run from July to December 2009, reports higher employment and earnings over an eight-year follow-up.

Gap: our checked sources don't show whether private, paid job search programs work as well as the evaluated programs above.

Gap: our checked sources don't test this help on people searching while still employed.

If you are over 50

The results are mixed.

  • A networking training made people about 12% more likely to leave unemployment insurance, with no evidence of an effect on the quality of their first job afterward (a Dutch randomized experiment with about 50,000 unemployed workers aged 50 and older, 2016 paper).
  • 72% of participants in a coaching program for older job seekers found a job, against 63% of the control group (a randomized field experiment by Patrick Arni, page read October 2026). In Arni's experiment, the gain was driven by people lowering the lowest pay they would accept (their reservation wage) and by searching more efficiently.
  • Across 82 impact estimates for unemployed people over 50, active labor market policies had on average a slightly negative effect on finding a job (0.8 percentage points) that disappears after 24 months (a meta-evaluation published on Cambridge University Press's site, page read October 2026). In that meta-evaluation, training had a positive effect: 2.4 points alone, 1.7 combined with search assistance or counseling. That meta-evaluation covers a wider set of policies than the Dutch and Arni experiments, so its average and their results may not conflict.

The counts of hires we hold do not support it. A Payscale article titled "How Many Jobs Are Found Through Networking, Really?" says some estimate upwards of 85 percent of open positions are filled through networking. Fox Business quotes recruiting consultant Lou Adler saying "85 percent of critical jobs are filled via networking of some sort". Fox Business says this is based on a survey of 3,000 people, from unemployed active job seekers to hiring managers.

The figures we hold differ by who was asked and what:

  • About half of US jobs are found via networks, according to Topa (2011), as cited in an NBER working paper on a randomized trial at a grocery chain.
  • Of 1,000 US workers, 54% said they had landed a job through a personal or professional connection (a May 2025 survey by the resume builder platform MyPerfectResume). In that May 2025 MyPerfectResume survey, 32% credited personal connections and 28% professional connections as most helpful in their most recent job search, ahead of job boards (13%) and staffing firms (8%).

None of these figures measures 85% of all hiring.

Where sources disagree

  • Do online applications work? Applications people sent on their own were the top source of hires from 2021 to 2025 (Ashby, a vendor of applicant tracking software, data on nearly 250,000 hires). But in survey panel data on people aged 18 to 64 from 2013 to 2017, applying to jobs (online or otherwise) was negatively associated with offers, while contacting an employer directly was associated with a higher offer rate (a supplemental appendix on the New York Fed's site). The authors of that New York Fed appendix flag that the people who choose each method may differ. The two may not conflict: one counts where hires came from, the other compares searchers who chose different methods.
  • How many posted jobs are real? In any given quarter, 18% to 22% of the jobs posted on Greenhouse's own platform are classified as ghost jobs, meaning positions advertised with no intent to hire (Greenhouse, a vendor of hiring software, 2024 State of Job Hunting report). Ashby, a vendor of applicant tracking software, says that from its customer base 18% of jobs open in 2024 were not filled, which is a different measure. There are no official statistics on how many ghost jobs exist (an April 2025 In Focus report titled "Ghost" Job Postings). The same April 2025 report says the firms behind the industry surveys it reviewed market hiring-related services and released varying levels of detail on their methods, making their estimates hard to assess.

Questions people also ask

How can you find a job quickly?

Gap: our checked sources don't rank ways of searching by how fast they lead to a job. Searching harder was related to more interviews, job offers and being employed, though the links were modest (a 2021 meta-analysis of 378 samples covering 165,933 people). In that 2021 meta-analysis, searching harder did not predict the quality of the job.

Gap: our checked sources hold no tested evidence on which first step works best. The closest test we hold is a randomized study of 1,097 unemployed young people in South Africa (described by J-PAL and in a working paper). In that South African study, those who also completed a detailed job search plan sent about 15% more applications than those who only attended a workshop, with no more hours spent searching. That South African study reports 30% more job offers and 26% more employment.

Users with tailored resumes reached an interview or offer at 5.75%, against 2.68% for those without (Huntr, a vendor, Q2 2025 report on over 1.39 million applications with self-reported outcomes). This Huntr figure is not a controlled test.

Showing job seekers relevant alternative occupations raised their total job interviews by 44% (an experiment by Belot, Kircher and Muller, reported in a July 2016 IZA discussion paper). In that 2016 paper, the 44% rise was a weak statistical result (significant only at the 10% level). In the same experiment, interviews doubled for narrow searchers compared with similarly narrow searchers in the control group. What Works Growth's summary of the same experiment reports significantly more job offers from people's other search activity. It also reports a negative and statistically insignificant effect on job finding (too small to tell from chance), in samples it calls quite small because of attrition (people dropping out of the study).

How we checked this: this page is the output of a rigorous research pipeline run with AI tools: every factual sentence rests on a finding checked against the source's own words, and a person approved the page before it went up. How we check our facts. Updated November 12, 2026. This page is general information, not legal, tax or financial advice.

Sources