Resume writing help raised hires by 8% and wages by 10% in a huge field experiment

In an experiment with nearly half a million jobseekers on an online labor market, those offered algorithmic writing assistance on their resumes were hired 8% more often at 10% higher wages, with no drop in employer satisfaction.

Share this:

Cleaner resumes get people hired. Jobseekers who received automatic writing help on their resumes were hired 8% more often and earned 10% higher wages, and the employers who hired them were no less satisfied.

The study is by Emma Wiles, Zanele Munyikwa and John Horton of MIT. It was published online on April 18, 2025, in Management Science, after first circulating as an NBER working paper.

The starting point was a correlation the authors saw in the platform’s own data: among new entrants, better-written resumes went with a higher chance of being hired, even after accounting for obvious differences between workers. The experiment was designed to test whether better writing actually causes more hiring.

Nearly half a million jobseekers

The experiment ran on a large online labor market. New jobseekers were intercepted as they wrote their resumes during registration, for a total sample of 480,948 people.

A randomly chosen treatment group was offered an algorithmic writing service that flags common errors and suggests fixes. The tool was nongenerative: it corrected writing rather than producing new text. The control group wrote their resumes with no assistance.

The authors then followed both groups through job search, hiring and the reviews written after jobs ended.

Fewer errors, more offers

Treated resumes had fewer grammar errors, redundancies and commonly confused words, and they were easier to read. The gains were largest for jobseekers whose writing started out weakest.

Importantly, the help did not change how people searched. Treated jobseekers sent no more applications and asked for no higher wages than the control group.

Even so, they were 8% more likely to be hired within their first month on the platform. Those who were hired earned hourly wages 10% higher, which the authors attribute to a shift in which workers got hired.

Employers were not fooled

A common worry is that polished writing hides a useful signal, so employers might hire weaker workers and regret it. The data did not show that.

Employer ratings at the end of contracts showed no significant difference between the groups. Treated workers were hired for at least as many hours and were just as likely to be rehired.

That result matters because the treated group was also paid more. If employers had simply been tricked into hiring worse workers, higher wages should have made disappointment even more likely to show up in the ratings.

The authors call this the clarity view, as opposed to a signaling view. Writing quality does not reveal ability so much as help employers see the ability a candidate already has, and they back this with a simple model in which writing help lowers the cost of producing a clear resume.

They note that Horton and Wiles have consulted for the marketplace where the experiment ran. Their practical conclusion is that digital platforms could build this kind of nongenerative writing assistance into how workers describe their services.

Study Details:

  • Title: Algorithmic Writing Assistance on Jobseekers’ Resumes Increases Hires
  • Authors: Emma Wiles, Zanele Munyikwa, John Horton
  • Journal: Management Science, Volume 71
  • Publication Date: April 18, 2025
  • DOI: 10.1287/mnsc.2024.04528