Every few weeks another version of the same post goes round: master these five, or seven, or ten social skills and watch your career take off. The advice is usually a mix of the obvious and the unfalsifiable, but the premise underneath it is not wrong. Social skills really have become more valuable at work. The useful question is what the evidence actually supports, because the honest version is both more interesting and considerably narrower than the promise.
The strongest data on this does not come from a productivity guru. It comes from labour economics.
What the labour-market data actually shows
In 2017, the economist David Deming published a paper in the Quarterly Journal of Economics called The Growing Importance of Social Skills in the Labor Market. Using US occupation data, he tracked how the mix of jobs changed between 1980 and 2012.
The jobs that require a lot of social interaction grew by close to 12 percentage points as a share of the American workforce over that period. Over the same years, jobs that were maths-intensive but less social, which includes many technical and STEM roles, shrank by a little over 3 percentage points. The market was tilting, measurably, toward work that involves dealing with other people.
That much fits the LinkedIn version. The part that gets left out is what grew fastest of all.
Why “soft skills” is a misleading name
In Deming’s data, the strongest growth in employment and pay was not in jobs that needed social skills instead of technical ones. It was in jobs that needed both. Social skill and cognitive skill behaved as complements, each making the other more valuable, rather than as substitutes.
His explanation is worth understanding because it is more concrete than “people skills matter.” Deming models teams as groups of workers who trade tasks to play to their strengths. Social skills, in this account, lower the cost of that coordination. They let a team divide work, hand parts back and forth, and adjust without everything grinding on friction. The value is not charisma. It is the reduction of coordination costs inside a group of people trying to produce something together.
This reframes the whole “soft skills” category. The term implies something gentle and secondary, a nice-to-have layered on top of the real, hard work. What the data suggests is closer to the opposite. In a workplace built on specialisation and teams, the ability to coordinate with others is part of the productive machinery, not a garnish on it.
It is worth being clear about what this evidence is and is not. Deming’s paper describes shifts across the whole US labour market at the level of occupations, measured over three decades. It documents where demand moved, not a promise about any single career, and its task measures are estimates rather than direct readings of what each worker does all day. A market-wide trend is a useful backdrop for a decision, not an instruction that everyone should retrain toward client-facing work. The direction of travel is well evidenced. The personal conclusion is yours to draw.
The payoff depends on the job
Here is where the self-help framing starts to overreach. “Master these skills and succeed” implies the return is universal, the same for everyone in every role. The research on the most popular of these constructs, emotional intelligence, suggests it is not.
In a 2010 meta-analysis in the Journal of Applied Psychology, Dana Joseph and Daniel Newman pulled together the studies linking emotional intelligence to job performance. Their integrative review found that the relationship was not consistent across the board. Emotional intelligence tended to predict better performance in jobs with high emotional demands, the roles heavy on dealing with people, and predicted performance poorly, sometimes even negatively, in jobs with low emotional demands.
That is a meta-analysis of many studies rather than a single result, and it points at something the listicles skip. The payoff from these skills is contingent. For a support lead, a salesperson, or a manager, emotional skill is close to the core of the job. For a role that is mostly solitary technical work, pouring effort into emotional finesse may add little and can distract from what the job actually rewards. Matching the skill to the work matters more than maxing out a generic trait.
What this means for the actual skills
None of this argues against getting better with people. It argues for being precise about which skills, and why. The evidence points less toward a personality overhaul and more toward a small set of learnable behaviours that reduce friction in a team.
Coordinating clearly is the central one, since that is the mechanism Deming’s model identifies: being easy to hand work to and take work from, saying what you will do and doing it, flagging problems early rather than late. Writing a message someone can act on without a follow-up call is a social skill in this sense, even though it looks like a technical one. So is listening closely enough to catch what a colleague actually needs, rather than what you assumed they meant.
These are duller than the usual advice about presence and influence, and that dullness is the point. They are specific, they are trainable, and their value shows up in whether a team runs smoothly, which is exactly what the economic data rewards.
It is also worth being honest that some of what gets sold as social skill is really about power and context. The same behaviour reads as confident from one person and abrasive from another, and workplaces are not neutral about who is allowed to be direct. A list of skills to master tends to quietly assume the playing field is level. It often is not, and no amount of active listening changes that on its own.
The reliable takeaway is smaller than the headline but more usable. Social skills genuinely matter more than they did a generation ago, they matter most alongside technical ability rather than instead of it, and their return depends on the job you are actually in. That is a case for developing a few concrete habits with some care, not for treating a viral list as a ladder.
