Every company says its people are its greatest asset, and most of them lose those people at a rate that makes the claim hard to believe. Replacing a skilled employee costs a substantial multiple of their salary once recruitment fees, onboarding time, lost productivity and the drag on colleagues are counted.
The question of what actually reduces turnover has been studied extensively, and the findings are less intuitive than the standard playbook suggests.
Pay matters, but not the way people assume
Compensation operates asymmetrically. Underpayment reliably drives people out; competitive pay does not reliably keep them in.
Once someone is paid at or slightly above market, additional increments produce diminishing returns on retention. The employee who receives a raise and stays is often the employee who was going to stay anyway.
This has a practical implication that many organisations resist: the counteroffer. Matching an external offer resolves the immediate departure in a minority of cases and rarely resolves the underlying reason the person started looking.
Where pay does matter enormously is perceived fairness. Internal inequity, particularly when new hires arrive on higher salaries than existing staff doing the same work, is one of the most reliable predictors of resignation.
The manager is the largest single variable
The most consistent finding across the research is that people leave managers more than they leave companies.
What specifically matters is narrower than general likeability. It comes down to three things: whether the manager gives clear expectations, whether they provide useful feedback rather than annual ritual, and whether they advocate for their team internally.
The organisational implication is uncomfortable. Most companies promote people into management based on individual performance and then provide no training in the actual job, which is a different job. The cost of that decision shows up in turnover statistics attributed to other causes.
Recognition: the part that is usually done badly
Recognition programmes are widespread and mostly ineffective, for reasons that are well understood.
They fail when they are predictable, when they are distributed for tenure rather than contribution, when they are perceived as arbitrary, and when they substitute for compensation rather than supplementing it.
They work when they are specific about what is being recognised, when they arrive close to the achievement, and when they carry social weight rather than just monetary value.
This last point explains why experiential recognition has held up better than cash bonuses in the evidence. A bonus is absorbed into household finances within a pay cycle; a shared experience is remembered and discussed. Programmes built around incentive travels, where teams that hit defined targets earn a trip together, combine three effects at once: the reward itself, the visibility of having earned it, and the relationship building that comes from time spent outside the office.
The honest caveat is that these programmes only work when the qualifying criteria are transparent and achievable. When people believe the outcome was decided in advance, the effect reverses.
Career visibility beats career progression
Employees do not primarily leave because they were not promoted. They leave because they could not see how promotion works.
Opaque progression, where advancement appears to depend on relationships rather than criteria, produces exit behaviour even in organisations with genuinely good mobility. What reduces it is publishing the criteria, showing examples of people who moved, and having managers discuss trajectory in a structured way rather than annually.
This is one of the cheapest retention interventions available and one of the least implemented, because it requires admitting how decisions are actually made.
Flexibility is now a hygiene factor
Working arrangements have moved from a differentiator to a baseline expectation in most knowledge roles.
The nuance is that flexibility does not mean the absence of structure. The arrangements that produce the best retention outcomes are those with clear, stable and predictable rules, whatever those rules are. Instability, where policy changes repeatedly, damages retention more than any particular policy.
What to do with this
For an organisation trying to reduce turnover with limited resources, the sequence that produces the most improvement per unit of effort is fairly consistent.
Fix internal pay inequity first, because it is corrosive and generates resentment beyond the individuals affected. Train managers second, because it is the largest single variable. Make progression criteria visible third, because it is nearly free. Redesign recognition fourth, because badly designed programmes are worse than none.
Exit interviews, worth noting, are the least reliable input in this whole exercise. People leaving are managing a reference and rarely tell you the real reason. Stay interviews with people who are not leaving produce considerably more honest information, and almost nobody conducts them.
