The free lunch was never free, but the price was cleverly denominated. An employee who eats three meals on campus, does laundry on campus, sees the dentist on campus and naps in a pod on campus has not received a gift. She has had the logistics of her life relocated inside the company's walls, which means the walls now contain her life. The medieval castle worked on the same principle: provision everything within the perimeter, and the perimeter stops feeling like one.
This is not a conspiracy theory; it is a design brief. Perks solve real problems for employers: they extend the workable day, they make the office the path of least resistance, and they create a lifestyle gradient between this employer and the next one. The employee who leaves does not just change jobs. She loses a cafeteria, a gym, a shuttle, a social world. Each perk is small; the switching cost is cumulative. That cumulative cost is the moat.
What perks buy, and from whom
It helps to ask what problem each perk solves and for whom. Free dinner solves the employee's dinner problem, but only if she is still at the office at dinner time; it solves the employer's problem of her leaving at six. On-site everything solves convenience, and also removes the daily errands that would otherwise take her off campus and remind her that a world exists out there. The perk portfolio is, in aggregate, an architecture of presence.
The deeper mechanism is psychological, and it is well documented. A meta-analysis of 128 experiments found that tangible rewards layered onto intrinsically interesting activities tend to undermine subsequent intrinsic motivation (Deci, Koestner and Ryan, 1999). The original experiments showed people paid to solve an engaging puzzle chose to play with it less afterward (Deci, 1971). The relevance to perks is direct: when a company answers the question "why do you work here?" with an inventory of amenities, it teaches people to answer that way too. The work itself quietly moves down the list of reasons.
And the amenity arms race has not, on the available evidence, produced an engaged workforce. Gallup's global workplace survey put employee engagement at 20 percent in 2025, the lowest reading since 2020, in a decade when workplace amenities have never been richer (Gallup, 2025). Correlation is not cause, but the coexistence is telling: whatever engagement is made of, it is apparently not made of snacks.
The benefit nobody dares to use
There is a second, quieter failure mode: the perk that exists on paper but not in behavior. As Ricardo Semler has told it, Semco in the 1980s gave every new father two weeks of paid paternity leave. Not one took it; each was afraid of what the boss would think. His conclusion became a rule: a benefit that nobody dares to use is not a benefit, it is decoration. The modern analogue is unlimited vacation, a policy that removes the countable entitlement and replaces it with a social calculation every employee must run alone, under the gaze of her manager.
Semco's response is the instructive part. Rather than exhort people to rest, it made vacation compulsory: thirty days, out of the building, laptop locked away. Semler's reading of the adults who tried to negotiate their way out of resting was blunt: that is not loyalty, it is fear, and fear is the cheapest management tool ever invented. A company that must force its people to use a benefit has learned something true about its own culture that no engagement survey would have surfaced.
The test for any perk, then, is not what it costs the company but what it costs the employee to use it fully. Parental leave that stalls careers, vacation that accumulates guilt, sabbaticals taken by nobody senior: these are moats in benefit costume, and employees learn to read them fluently even when the employer brand does not.
Amenity or salary, presence or output
None of this argues for stripping offices to folding chairs. Good food and good spaces are pleasant, and pleasantness is not a sin. The question is what the amenities are instrumented to do. There is a version of the well-provisioned workplace that shortens the day, and a version that lengthens it; a version that assumes adults will go home, and a version engineered so they will not want to.
A useful audit takes one afternoon. List every perk and ask: does this make it easier to leave at a reasonable hour, or harder? Does it substitute for cash the employee could spend on a life of her own design, or add to it? Would a manager notice, and mind, if it were fully used? Perks that fail the audit are not compensation. They are the moat, and the moat, whatever else it does, was never dug for the benefit of the people inside.
Sources
- Deci, Koestner & Ryan (1999). A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation. Across 128 experiments, tangible rewards undermined subsequent intrinsic motivation for interesting tasks. https://doi.org/10.1037/0033-2909.125.6.627
- Deci (1971). Effects of Externally Mediated Rewards on Intrinsic Motivation. Subjects paid to solve an engaging puzzle later chose to engage with it less, the original demonstration of the undermining effect. https://doi.org/10.1037/h0030644
- Gallup (2025). State of the Global Workplace. Global employee engagement fell to 20% in 2025, the lowest level since 2020. https://www.gallup.com/workplace/659610/state-global-workplace.aspx