Return-to-office mandates are often sold as a rescue plan for junior talent. Put people back together and mentoring will flourish, the story goes. The evidence points elsewhere. Mentoring is generally beneficial, but the gains are measured in inches, not miles, and they depend on what you are trying to improve. Couple that with an aging workforce and a widening training gap among older employees, and the question is not where people sit. It is how you design learning across ages.
What the research actually says
A seminal meta-analysis in the Journal of Applied Psychology found that mentored employees fare better than their non-mentored peers on multiple outcomes, yet the effects on hard career metrics are small (Allen et al., 2004). The association with compensation was weak, roughly r = .12, and the link to promotions was modest, around r = .31. The strongest benefits showed up in subjective outcomes such as career satisfaction. In other words, mentoring nudges objective progress and more reliably boosts how people feel about their work and trajectory.
A broader meta-analysis that pooled 112 studies across youth, academic, and workplace contexts reached a similar conclusion: mentoring correlates with favorable outcomes across behavioral, attitudinal, health, relational, motivational, and career domains, but the typical effect is small (Eby et al., 2008). For example, corrected correlations were about .14 for career attitudes and .16 for situational satisfaction. The context matters too. Academic mentoring showed the largest associations. Workplace mentoring sat in the middle, with effects commonly in the .03 to .19 range. The authors advise against overselling mentoring, a useful check for firms tempted to promise promotions and pay bumps on the back of a few coffee chats.
Read plainly, these findings do not diminish mentoring. They right-size it. If you expect a program to move pay scales or promotion rates in a quarter, you will be disappointed. If your aim is to increase satisfaction, commitment, and the flow of know-how, mentoring is a credible tool. That is a design question, not a seating chart.
The demographic shove toward cross-age learning
This is not an academic debate. The age structure of the workforce is changing fast. The old-age dependency ratio in the OECD, the population aged 65 and over relative to the working-age population, has risen from 19 percent in 1980 to 31 percent in 2023. It is projected to reach 52 percent by 2060 (OECD, 2025). Without productivity gains or policy changes, this demographic shift could slow average OECD per capita GDP growth by about 40 percent, from 1 percent a year in 2006–19 to 0.6 percent in 2024–60. Put bluntly, economies will depend more on older workers staying productive for longer.
Yet older workers are trained less. Only about one third of adults aged 60 to 65 participated in training in 2023, compared with more than half of those aged 25 to 44 (OECD, 2025). Many older adults also face lower digital literacy, compounding the risk that critical tacit knowledge does not travel to younger colleagues while new tools do not travel back.
That makes intergenerational learning a core operating issue. Two-way mentoring is an obvious candidate, but here the evidence base is thinner. The multidisciplinary meta-analysis flags gaps in research on reverse mentoring and on age-diverse mentoring triads, so firms should treat these designs as promising but still under-studied, then measure their effects locally (Eby et al., 2008). The demographic pressure is clear. The exact mentoring mechanism that best transfers knowledge across ages is still a strategic choice that demands testing.
How to design mentoring that pays its way
Treat mentoring like any other management system: define the outcome, choose the mechanism that most directly produces it, and measure what you get.
-
Aim at outcomes mentoring reliably moves. The strongest, most consistent gains are in subjective domains such as career attitudes and situational satisfaction (Eby et al., 2008), and in overall career satisfaction for protégés (Allen et al., 2004). If retention, engagement, or readiness for stretch assignments are your near-term goals, spell that out and track them.
-
Be honest about hard metrics and the time horizon. The association with compensation is small and promotions are only modestly linked to mentoring (Allen et al., 2004). If you expect quick changes in pay or title, you will either game the metric or give up. Use longer windows for objective outcomes and balance the scorecard with near-term indicators mentoring is more likely to move.
-
Match by work, not by birth year. The workplace effects are smaller than in academic settings (Eby et al., 2008), which suggests that context and task fit matter. Pair people around concrete problems, projects, or skills, not stereotypes about generations. The more the exchange is anchored in real work, the less it relies on charisma or happenstance.
-
Protect time and make progress visible. Small average effects often reflect inconsistent exposure. Put mentorship time on calendars, keep it brief, and ask pairs to record goals and outcomes. When people can see momentum, the subjective benefits that mentoring is good at improving will show up in your data.
-
Make it two-way, then test and learn. Older workers hold institutional memory and client nuance, younger workers may have newer tools and methods. Reverse mentoring is a plausible design to bridge that gap, but the research base is limited, so treat it as a hypothesis and evaluate it with the same discipline you bring to a product rollout (Eby et al., 2008).
-
Pair mentoring with formal training for older workers. The participation gap in training among adults aged 60 to 65 is large (OECD, 2025). Mentoring will not fill a basic skills deficit on its own. Where digital fluency or new systems are the constraint, add short, targeted training. Use mentoring to contextualize and reinforce the learning in live work.
The practical read
The strongest evidence says mentoring is worth doing, but not as a silver bullet. It reliably improves how people feel about their careers and their connection to the organization, and it can modestly nudge promotions and other objective outcomes over time (Allen et al., 2004; Eby et al., 2008). Meanwhile, the age mix of the labor force is shifting toward more older workers, and training participation among those workers lags, raising the cost of wasted knowledge and stalled upskilling (OECD, 2025). The smart response is not to herd everyone back to the same building and hope. It is to design intergenerational mentoring like a system: explicit goals, work-based matches, protected time, two-way exchange, and measurement.
Do that, and mentoring becomes a practical part of how your company renews itself. Do anything less, and you are betting on chance meetings to solve an aging problem.
Sources
- Allen, Eby, Poteet, Lentz and Lima (2004). Career benefits associated with mentoring for protégés: a meta-analysis. Shows mentoring is linked to positive career outcomes, with small effects on compensation and modest links to promotion, and stronger effects on subjective satisfaction. https://www.semanticscholar.org/paper/Career-benefits-associated-with-mentoring-for-a-Allen-Eby/a13c80a7f263f4378c07b474a0628105b91c6d28
- Eby, Allen, Evans, Ng and DuBois (2008). Does Mentoring Matter? A Multidisciplinary Meta-Analysis. Synthesizes 112 studies and finds generally small effects across domains, with workplace mentoring effects typically .03 to .19, and cautions against overselling; notes gaps on reverse mentoring. https://pmc.ncbi.nlm.nih.gov/articles/PMC2352144/
- OECD (2025). OECD Employment Outlook 2025. Documents rapid population aging, a projected 40 percent slowdown in per capita GDP growth absent productivity gains, and lower training participation among adults aged 60–65, making cross-age learning a structural need. https://www.oecd.org/en/publications/oecd-employment-outlook-2025_194a947b-en.html