A KPMG survey of 1,019 professionals found that 57 percent would choose a job paying 10 percent below market with close workplace friendships over one paying 10 percent above market without them—placing human connection above a 20-point difference in pay.

A job paying 10 percent above market appears, on paper, to have an obvious advantage over one paying 10 percent below. KPMG asked what happened when the lower-paying role came with close workplace friendships and the higher-paying one came without them. In its preference model, 57 percent chose the first.

The result gives human connection unusual weight. It also needs careful handling. Nobody in the study accepted a real offer, surrendered part of a salary or discovered whether promised friendships survived an ordinary Monday. Respondents evaluated hypothetical employers.

This is one corporate survey, not a peer-reviewed experiment or a settled exchange rate between friendship and pay. Read within that boundary, it still says something useful: many people judge a job partly by the relationships through which they will have to live it.

What KPMG actually measured

The KPMG Friends at Work 2.0 survey was fielded online from 17 to 30 June 2025 among 1,019 full-time professionals in the United States.

The public methodology describes them as employees in office-based roles requiring specific skills or higher education. They worked at organisations with at least 100 employees across a mix of industries. Job titles ran from entry level to senior vice-president; the C-suite was excluded.

KPMG presented respondents with a series of employer profiles. Five attributes changed: salary, workplace friendship, work-life balance, learning opportunities and company culture. Each could appear at a low, middle or high level.

For friendship, those levels ranged from no meaningful connections, through some friendly work relationships, to close workplace relationships. Salary ranged from 10 percent below market to market rate and 10 percent above. The model began with an average employer, with all five factors at their middle level, producing a 50 percent benchmark for preferred-employer status.

KPMG then modelled an employer offering close work friendships and salary 10 percent below market, while keeping work-life balance, learning and culture at their average levels. Its predicted preference rose to 57 percent. The comparison on the other side was a role paying 10 percent above market without close friendships.

The finding is hypothetical, but not meaningless

KPMG describes this as workers valuing friendship at a 20 percent salary premium. A more precise description is that the two salary levels sit 20 percentage points apart relative to the market benchmark: 90 percent of market versus 110 percent.

That wording matters because a predicted preference is not the same as observed behaviour. Nobody had to compare actual benefits, commuting costs, contract security, childcare, promotion prospects or the credibility of a manager promising close friends. Nobody had to live on the lower number.

Hypothetical choices can reveal priorities because the trade-off forces one attribute into view. They can also make a difficult choice cleaner than life permits. A respondent can choose friendship in a survey without confronting what a 10 percent reduction would remove from a household budget.

The finding therefore supports a narrower statement than “people will give up 20 percent of their pay for friends”. It shows that, in KPMG’s constructed employer profiles, close relationships shifted preference enough to outweigh a large stated salary difference for a modest majority of this sample.

Why friendship enters the economics of a job

A salary is paid for time, but the social quality of that time is part of what the employee receives. A trusted colleague can make a difficult project less isolating, supply context that never reaches a formal memo and turn a day of transactions into a day containing some ordinary recognition.

The rest of KPMG’s survey helps explain why friendship affected the hypothetical choice. Eighty-seven percent rated the ability to have close work friendships as very important, using the top two points on a seven-point scale. Twenty-eight percent selected increased productivity and motivation as one of the three largest professional benefits they attributed to close work friends.

Those are perceptions, not independently measured gains. They nevertheless show that respondents did not treat workplace friendship as something that begins only after work ends. They connected it to information, confidence, problem-solving, opportunity and the experience of doing the work itself.

Money also shaped whether those relationships could develop outside the office. Three quarters said their finances or the economy inhibited social meetings with colleagues to some extent, up from 54 percent in KPMG’s 2024 survey. Even the supposedly non-financial side of work has a financial cost when friendship requires coffee, dinner, travel or time that could have been used elsewhere.

A company can offer proximity, not friendship

There is an important asymmetry inside the hypothetical profile. An employer can put a salary in a contract. It can create offices, shared projects and time for people to meet. It cannot guarantee that one employee will find another person they trust.

A recent Silicon Canals article examined University of Kansas research suggesting that close friendship takes more than 200 hours and depends especially on voluntarily chosen time. Work was one of the main places adults met potential friends, but obligatory time together was less strongly associated with closeness than time people chose to share.

That distinction puts a limit on the 57 percent scenario. “Close workplace relationships” appeared as a finished attribute, like salary or culture. In real life it is an uncertain outcome produced by two people, over time, with no obligation to cooperate with an employer’s retention plan.

A company can improve the odds by leaving room for informal conversation, avoiding schedules that make every interaction instrumental and allowing social time to remain voluntary. It can also damage the odds by turning friendship into a managed performance requirement. Connection becomes less credible when attendance is compulsory and managers are counting.

Earlier Silicon Canals coverage found a similarly demanding signal

Silicon Canals has previously looked at Gallup’s controversial “best friend at work” engagement item. The word “best” often makes executives uncomfortable because it sounds too personal for an employee survey. Gallup says that softening it to “friend” or “good friend” reduced its ability to distinguish highly productive workgroups.

That finding and KPMG’s salary scenario are not interchangeable. Gallup’s item concerns an association with workgroup outcomes; KPMG modelled employer preference. Neither establishes that friendship causes performance or that organisations can produce friendship on demand.

Both do point to the same overlooked distinction. Collegiality and closeness are not synonyms. A workplace can be polite, efficient and broadly supportive while offering nobody the kind of relationship in which a person can speak plainly, ask for help or be understood without restating the entire background.

KPMG’s 2026 follow-up complicates the retention story

KPMG’s 2026 Friends at Work study, conducted with CivicScience, makes the organisational story less convenient. Employees with close personal friends at work were the most engaged group in that study. They were also the most likely to say they were searching for another role: 42 percent said they were very likely to job-search within the next 12 months.

The 2026 survey used different samples and a different design, with sample sizes varying by question, so it cannot simply be laid over the 2025 result as a trend line. Its relationship categories were also more specific, separating close personal friends, workplace friends, collegial relationships and people who did not socialise at work.

Still, it challenges the easiest managerial conclusion. Close friends may make someone engaged without making them captive. A strong workplace relationship can supply confidence, contacts and information about other opportunities. The friend may even leave first.

The survey does not authorise a friendship discount

The most cynical reading of the 57 percent figure would be that employers can pay less if the office feels socially warm. The survey does not establish that.

Work-life balance, not friendship, was the strongest driver of job preference across KPMG’s five attributes. A profile combining strong work-life balance with frequent, high-quality learning opportunities reached 71 percent preference in the model. Respondents were weighing bundles of conditions, not declaring compensation irrelevant.

Nor did the study examine what happens when below-market pay is sustained for years, compounds into lower pension contributions or creates resentment as workers compare salaries. A hypothetical role with guaranteed friends removes many of the uncertainties that make actual underpayment corrosive.

Employers can take the social result seriously without converting it into a discount. Fair compensation answers one part of the employment relationship. Time, trust, autonomy and connection answer others. One does not cancel the obligation to think about the rest.

There is also a fairness question around who gets included. Close workplace circles can support their members while leaving newcomers, contractors, remote staff or anyone outside the dominant social group with less information and fewer advocates. A friendship-friendly workplace is not automatically an inclusive one.

The number exposes what a salary-only model leaves out

KPMG’s public deck gives broad sampling criteria, but it does not provide enough detail to treat the 57 percent as a precise estimate for every profession, country or type of work. The sample was American, full-time, office-based and drawn from organisations with at least 100 employees. Its answers arrived during a particular labour market in June 2025.

The figure is still useful because it makes an often hidden part of job choice visible. People do not experience employment as a payslip beside an empty desk. They experience managers, confidants, interruptions, lunches, difficult conversations and the relief of not having to explain everything alone.

Human connection ranked above a wide hypothetical salary gap for 57 percent of this sample. That does not put a stable price on friendship. It shows the limits of pretending salary is the only thing workers are choosing.