The Digital Whale

Social media and communication

Whether an online presence changes hiring

There is no controlled evidence that personal branding works, and the measurable effects of an online presence on callbacks run through discrimination.

The honest summary is negative. Not one controlled study tests whether personal branding — posting consistently, holding a niche, publishing thought leadership — improves hiring or earnings outcomes, so the advice industry built on it rests on nothing measured. The evidence that does exist comes from field experiments that sent real applications with randomly varied online profiles, and what those experiments detect is discrimination: candidates whose profiles signal a minority religion or a foreign origin are called back far less often, and profile photographs move interview rates on their own. Recruiters' judgements of profiles, separately, do not predict who will do the job well. The measurable effect of an online presence is mostly something done to candidates rather than something they manage. For a workplace-oriented example of technology measurement in practice, see top remote companies.

Employers say they check far more than they do

The number underpinning most screening advice comes from a survey commissioned by a company selling recruitment services.

2017

70% of US employers reported using social networking sites to research candidates, 54% said they had found content that stopped them hiring someone and 44% content that led them to hire someone.

Direction: Increase. Strength of evidence: Weak.

CareerBuilder / Harris Poll, 2017Opt-in online survey of 2,380 US HR managers

Caveat Commissioned by a job board with a commercial interest in the finding, self-selected, and nine years old; self-reported screening is heavily inflated against observed behaviour.

A field experiment that sent thousands of real applications and watched for search activity put the true rate at a fraction of that.

2020

Only an estimated 10–33% of US employers actually searched for candidates online, far below the roughly 70% employers report in surveys.

Direction: No detectable effect. Strength of evidence: Strong.

Alessandro Acquisti & Christina Fong, Management Science, 20204,173 real job applications to US employers

Caveat Fieldwork was in 2013 on entry-level US roles, and search behaviour is inferred from callback patterns and web traffic rather than observed directly.

The other survey the advice industry leans on is SHRM's 2025 Talent Trends, which reported that 55% of organisations had used social media to reach candidates in the past year — the most-used recruiting method, though social media did not appear among the methods those same employers rated most effective. It is a self-selected sample of 2,040 HR professionals from a professional association's own panel, not a probability sample of employers.

What a profile reliably changes is the effect of identity cues

The same field experiment varied one thing in each candidate's online presence and counted callbacks. Sexual orientation produced nothing; religion produced one of the larger discrimination effects in the literature.

2020

A candidate signalling a Muslim identity received a 6.25% callback rate in Republican-leaning counties against 22.58% for an otherwise identical Christian candidate, while nationally and for the gay-against-straight comparison (10.69% against 10.63%) there was no significant difference.

Direction: Decrease. Strength of evidence: Strong.

Acquisti & Fong, Management Science, 20204,173 US applications with randomly assigned online identity cues

Caveat Employer political leaning was proxied by county-level data rather than measured on individual hiring managers, and the subgroup cells are small.

A French experiment produced a comparable result and then, by accident, a control condition: Facebook changed its layout mid-study in a way that made the origin cue less visible.

2019

Callback rates differed by 41.7% between two otherwise identical candidates whose Facebook profiles signalled French against Moroccan origin, and the gap vanished after a layout change reduced the visibility of that cue.

Direction: Decrease. Strength of evidence: Strong.

Manant, Pajak & Soulié, Journal of Economics & Management Strategy, 2019Over 800 applications over 12 months, two matched fictitious candidates

Caveat Two fictitious applicants in one country, fieldwork in 2012 on a platform since redesigned, so the natural experiment is powerful but cannot be replicated.

The gap closing when the cue became harder to see is the closest thing to a mechanism in this literature, and the strongest argument for understanding what a digital footprint exposes: the operative variable was visibility, not content.

Photographs move callbacks on their own

Appearance is the one profile element with a measured effect that is not about group identity, and no content strategy addresses it.

2018

Varying only the Facebook profile picture changed interview invitations by about 38% in relative terms, from a 5.3% callback rate for the least favourable photo to 9.8% for the most favourable.

Direction: Increase. Strength of evidence: Mixed.

Stijn Baert, New Media & Society, 20181,056 Belgian job applications with four randomly assigned profile photos

Caveat The design measures the relative effect of four photographs and cannot measure how often employers looked, nor the effect of having a profile at all.

Recruiter judgements of profiles predict nothing useful

If screening profiles worked as a selection tool, ratings of them would forecast performance. Tested directly, they do not, and they carry a demographic tilt.

2016

Recruiter ratings of applicants' Facebook profiles were essentially unrelated to supervisor-rated job performance (r = −.13 to −.04), turnover intentions (r = −.05 to .00) and actual turnover (r = −.01 to .01), added nothing beyond cognitive ability and personality tests, and favoured female and White applicants.

Direction: No detectable effect. Strength of evidence: Strong.

Van Iddekinge, Lanivich, Roth & Junco, Journal of Management, 2016College-aged applicants for entry-level roles, recruiter ratings matched to later outcomes

Caveat The null concerns predictive validity for the employer, which is a different question from whether a candidate's profile changes their odds of being called.

Networks move jobs, but not in the way the advice describes

The one large experimental result about professional platforms concerns connections rather than content: LinkedIn randomised its recommendation algorithm across years of user cohorts and tracked the job moves that followed.

2022

Across experiments covering more than 20 million users, tie strength and job mobility follow an inverted U: moderately weak ties produced the most job transmissions, and the strongest ties the fewest.

Direction: Increase. Strength of evidence: Strong.

Rajkumar, Saint-Jacques, Bojinov, Brynjolfsson & Aral, Science, 2022Randomised experiments on over 20 million LinkedIn users across five years

Caveat Two authors were LinkedIn employees with financial interests in Microsoft, the user base skews to finance and technology, and the analysis is intent-to-treat because users choose whether to act on recommendations.

That is a finding about the shape of a network, not about publishing. It says nothing about follower counts, posting frequency or tone. Advice about how to look for work online can lean on it only in the weak sense that acquaintances matter more than close contacts.

The short version

  • Employers report screening candidates online at roughly 70%, but a field experiment estimates the real rate at 10–33% (CareerBuilder, 2017; Acquisti & Fong, 2020).
  • The largest measured effect of an online profile is discriminatory: 6.25% against 22.58% callbacks for a Muslim against a Christian candidate in Republican-leaning US counties (Acquisti & Fong, 2020).
  • A French experiment found a 41.7% callback gap on origin cues that closed when a platform redesign hid the cue (Manant, Pajak & Soulié, 2019).
  • Recruiter ratings of profiles do not predict job performance or turnover and add nothing to standard selection tests (Van Iddekinge et al., 2016).
  • No controlled study has tested personal branding at all, so claims that it improves hiring outcomes are untested rather than merely contested.