Almost every large company has a set of stated values, an annual engagement survey and a programme of workshops intended to strengthen culture. Almost every large company also has employees who can describe, accurately and cynically, the gap between the stated values and how decisions actually get made. The gap is not hypocrisy. It is a design failure, and it has an identifiable cause.
Culture is the residue of decisions
Employees infer what an organisation values not from what it declares but from what it rewards. Who gets promoted, whose projects survive a budget cut, what behaviour is tolerated in a high performer, and how the company behaves when a value is expensive — these observations form culture faster than any communication programme. A company that lists integrity first and promotes a manager known for shading numbers has taught a lesson that no workshop will unteach.
Read moreThe Real Economics of Corporate Layoffs
This is why culture change initiatives that begin with language almost always stall. The language is the last thing to change, not the first. Organisational research on this sequencing is extensive, and reviews compiled by the Society for Human Resource Management consistently find that perceived fairness of promotion and pay decisions predicts engagement more strongly than any communication variable.
The measurement problem
Engagement surveys are useful and frequently misused. Aggregate scores reported upward tend to be smoothed, because a manager whose team scores poorly has an incentive to manage the survey rather than the team. Response rates and comment content usually carry more signal than the headline index, and the most informative question in most instruments is whether an employee would recommend the organisation to a friend — because it is difficult to answer strategically.
Attrition data, disaggregated by manager and by tenure, is harder to game and more actionable. Voluntary departures concentrated under a single leader, or a spike in departures shortly after a compensation cycle, describe a culture problem more precisely than a survey ever will.
Read moreWhy AI Data Centers Are Reshaping U.S. Power
Safety, ethics and the cost of silence
The most consequential cultural variable in operational businesses is whether employees report problems. Organisations where raising a concern is career-limiting accumulate hidden risk until it surfaces as an incident, a recall or an enforcement action. Investigations of major industrial and financial failures repeatedly identify prior internal warnings that did not travel upward, and enforcement guidance on compliance programme effectiveness published by the U.S. Department of Justice now explicitly asks whether employees who reported concerns were protected or penalised.
Whistleblower frameworks exist partly because internal channels fail so often. Award programmes administered by the Securities and Exchange Commission have paid out substantial sums to insiders whose internal reports were ignored, which is an expensive way for a company to discover its speak-up culture was theoretical.
What actually moves it
Three changes do measurable work. Make promotion criteria explicit and publish how they were applied. Give managers fewer direct reports so that development conversations can be real. And remove a high performer who violates a stated value, publicly enough that the organisation notices — the single most credible cultural signal available to any leadership team, and the one most consistently avoided.
Read moreThe Economics of the American Second Home
Where the money actually goes
Culture budgets tend to concentrate in the most measurable and least consequential activities. Survey platforms, offsite facilitation, values workshops, and internal communications campaigns all produce artefacts that can be counted, which makes them easy to fund and easy to report. What they do not touch is the allocation of scarce resources, the criteria used in promotion committees, or the behaviour tolerated in high performers who deliver revenue. Employees read those three things as the real culture and treat everything else as marketing. Research summarised by SHRM repeatedly finds that stated values predict behaviour poorly when incentive systems point elsewhere.
This is why the most credible culture interventions look unglamorous. Rewriting promotion rubrics, publishing the criteria by which pay bands are set, changing who sits in the room when headcount is allocated, and removing a manager whose numbers are good but whose attrition is catastrophic all send unambiguous signals. None of them require a workshop.
The measurement problem nobody solves
Culture programmes are also structurally hard to evaluate because the metrics chosen are usually attitudinal and self-reported, collected on the same instrument that the programme was designed to improve. Engagement scores rise after a communications push and drift back within two quarters, which is consistent with either a real but temporary effect or with pure response bias, and the typical measurement design cannot distinguish the two. Organisations that take this seriously anchor on behavioural outcomes instead: voluntary attrition among high performers, internal fill rates for senior roles, time from concern raised to concern resolved, and the proportion of promotions that come from outside the incumbent leader network. Those numbers move slowly, resist gaming, and are considerably less comfortable to present to a board than a rising engagement index.
