Where you sit in the ecosystem determines what the bargain looks like – and whether leadership still has the right to rewrite it
| KEY PROPOSITION
The rewards reset is not a single event. It is a contested renegotiation of give and get. What dies, what remains and what is born depends on whether you are employer, employee, shareholder, union, customer or policymaker – and on whether leaders treat the reset as a cost exercise or as a test of legitimacy. |
Do not panic. Do not be naïve. According to Ntombizone “Zone” Feni, Group CEO of 21st Century, the exchange between employer, employee and society is being reset in real time, in tighter budgets, sharper board questions, employees who want more than a salary increase, informal use of artificial intelligence at the desk, and a South African disclosure regime that has moved from polite King language to statutory consequence. The status quo still treats this as housekeeping: tidy the grades, refresh the brochure, add a wellbeing app. That assumes there is one bargain and one professional task. There is not. The reset looks different depending on where you sit. Miss that, and you will design for your own seat while the rest of the system withdraws consent.
The reset is not coming. It is already in the room
Reward no longer sits in HR alone, and it certainly does not sit in payroll alone. It sits in an ecosystem where employers, employees, unions, shareholders, customers, policymakers, and the wider economy collide. Each party gives something and expects something back. Reward works when every give has a fair get. Break the balance anywhere, and the whole system feels it.
That is why a “market adjustment” can satisfy finance and infuriate a high-impact specialist, and why a wellbeing campaign launched beside unpaid overtime reads as theatre. The bargain has been rewritten before. What is different now is speed, scale and simultaneity. Generation Z is not the cause; it is the stress test. Younger cohorts arrive with visible salary comparisons and tools that can challenge an expert in seconds. They will not trade present fairness for a ladder that AI may remove before they reach it (EY, 2025; Rudolph et al., 2021). Portfolio work is already a resilience strategy, even as the health costs of precarious work remain poorly designed for (Cropanzano et al., 2023; Margerison et al., 2025). Wellbeing is no longer a moral extra. Overloaded systems do not become productive because the posters improve (World Health Organization, 2022).
What dies, what remains, what is born
When systems reset, some things should die, some must endure, and some entirely new things emerge. That framing, set out by Ntombizone Feni (Group CEO, www.21century.co.za) for the profession, is the right starting point – provided we stop treating it as a single list that looks the same from every chair. What should die is blunt, one-size-fits-all design; hidden logic that cannot survive a town hall; benefits unused in real life; and symbolic wellbeing that leaves workload, debt stress, gender-based violence and male silence untouched. Those practices were never best practice. They were convenient practice.
What must remain are the unfashionable fundamentals: fairness, trust, growth, governance and human judgement. Decades of organisational justice research show that distributive, procedural, interpersonal and informational fairness shape trust, commitment, performance and withdrawal (Colquitt et al., 2001; Colquitt, 2023). People do not require every outcome to favour them. They do examine whether decisions are consistent, explanations truthful, and the powerful subject to the same rules. Fair process strengthens identification and voluntary cooperation because it signals that people are members of the group, not disposable inputs (Tyler and Blader, 2003).
What is being born is more demanding: targeted benefits that follow evidence of need; role-based flexibility instead of a single remote-work slogan; explainable pay decisions that can survive both Section 30B scrutiny and an employee who has already seen a comparator online; and AI-shaped notions of value in which output, judgement and tool mastery matter more than hours at a desk. Process transparency – how decisions are made – appears to matter more for satisfaction than dumping raw numbers into the market (Smit and Montag-Smit, 2025). What moves behaviour is transparency that reveals inequity, not mere inequality (Gutierrez, Obloj and Zenger, 2025; Cullen, 2024).
The view depends on the seat
This is the point the profession still underplays. There is no single reset. There are several, running at once.
From the employer seat, the reset looks like cost pressure, scarce specialist skill, disclosure risk and the demand to do more with a smaller budget. AI already changes work informally, whether policy has caught up or not. The rational fear is getting the call wrong in public.
From the employee seat, the reset looks like a bargain that no longer pays in the old currency. Loyalty is a weaker store of value. Cash still matters, but so do growth, flexibility and a credible answer to the question: if AI does the junior work, what am I here to become? Perceived breaches of the employment relationship create withdrawal even where no formal contract has been broken (Morrison and Robinson, 1997).
From the shareholder and board seat, the reset looks like governance. King V Principle 11 requires fair, responsible and transparent remuneration in service of sustainable value creation (Institute of Directors South Africa, 2025). Sections 30A and 30B of the Companies Act now give that language statutory teeth for public and state-owned companies: a binding policy vote, an annual implementation report, top-to-bottom five per cent pay-gap disclosure, and a two-strike rule that can bar non-executive remuneration-committee members from that committee for two years (Republic of South Africa, 2024). Boards cannot outsource judgement.
From the practitioner seat, the reset feels like being trapped between fairness, affordability, disclosure, talent demand and incomplete data. That discomfort is not a personal failing. It is the job becoming what it always claimed to be: strategic navigation rather than administration.
Customers and policymakers sit in the same ecosystem. A younger customer base will not indefinitely buy from firms whose internal bargain looks extractive. Policymakers are writing disclosure rules because market self-regulation was too slow. Treating either party as “external” to reward is how the profession stays late to its own argument.
The legitimacy test sitting underneath the numbers
A title gives authority. It does not give legitimacy. Authority can compel compliance with a new pay framework. Only legitimacy earns commitment to it. Organisational legitimacy is the judgement that actions are appropriate and consistent with accepted norms (Suchman, 1995). If executives retain concentrated reward while junior developmental work is automated away, if attendance rules apply downwards, if “the algorithm” is blamed for a decision no named human will defend, the organisation is not modernising. It is rebuilding the collapsed pyramid with better software. People may even tolerate harsher evaluation from an AI manager because the social checks that constrain extractive human bosses are muted (Dong et al., 2025). That is a governance hazard, not a feature. The economic case for leaner work is real: generative AI can raise speed and quality and lift less experienced workers fastest (Noy and Zhang, 2023; Brynjolfsson, Li and Raymond, 2025). Used as a democratiser of expertise, it widens access to method. Used as a digital foreman, it concentrates control and dissolves accountability. Gains that travel only upwards while insecurity travels downwards will be read as a breach, not as transformation.
A practical guide for companies: build a reward moat
A moat in reward is not a prettier benchmark pack. It is a protected evidence estate plus the judgement to use it. Confidential people data is entrusted capital, not a market commodity. Insight is the value. Trust is the licence to operate. Companies that want to get ahead of the reset, rather than wait for the next disclosure cycle to choose for them, should do five unglamorous things.
- Protect and connect the people-data estate. Pay numbers without job architecture and role context are orphan statistics. Connect remuneration to evaluated work and sector pattern, and you can explain a number rather than merely publish it.
- Build repeatable fairness diagnostics before the narrative. A repeatable baseline – band compliance, tenure against pay, race-in-grade, gender-in-grade – turns Section 30B and EEA4 from an annual scramble into owned findings, an exceptions register and a year-on-year line.
- Share productivity gains in the bargain, not only in the investor deck. If AI produces more output with fewer hours, some of the surplus must fund development, better work design and more meaningful roles. Otherwise the company structure stands on sand.
- Keep a named human on every consequential decision. AI may inform grading, shortlisting or incentive allocation. A named leader must own it, explain it and provide a route of appeal. Voice before deployment beats consolation afterwards.
- Measure capability creation as seriously as cost. A structure that improves this year’s margin while destroying next decade’s judgement pipeline is consuming its own future. Let AI raise speed; keep context, ethics and accountability human.
A practical guide for employees: safeguard yourself in the reset
Employees are not passengers in this ecosystem. The old advice – keep your head down, collect years, wait for the grade – is now a high-risk strategy. The safer posture is portable value, visible contribution and a refusal to outsource your own bargain.
- Build a portfolio of proof, not a portfolio of job titles. Document output, problems solved and judgement exercised. People who can show impact will negotiate from evidence. Those who can only show tenure will negotiate from nostalgia.
- Treat AI fluency as contribution, not cheating. The market has always rewarded people who produce more with better tools. Master them, then insist on being assessed for the reasoning you still have to defend. Prompt operators without judgement are the new clerical risk.
- Diversify income and capability without romanticising precarity. A side skill or structured independent work can be resilience. An uninsured, algorithm-managed grind is not freedom. Price the risk into the deal.
- Ask for process, not rumours. Request the logic of your band, the comparator group and the path to the next range. Internet gossip is a poor substitute for process transparency, and it often lowers satisfaction rather than raising it (Smit and Montag-Smit, 2025).
- Protect the human system you actually live in. Debt, burnout, caregiving and safety are not private hobbies outside reward. Use the benefits that have substance. Decline the theatre. Inclusion that tells men to be silent and women to wait their turn is just another broken bargain.
The choice in front of the profession
We can wait and see, adapt a little late, or get ahead and design the next fair exchange on purpose. Reward is moving from administration to navigation. The organisations that keep a licence to operate will protect confidential evidence, connect pay to the work beneath it, share the gains of new tools, and keep judgement attached to named leaders. The employees who keep options will show value, use tools without surrendering thinking, and refuse loyalty without reciprocity. There is no give and no get without people. If we lose that, no amount of disclosure will save a system that has already lost the right to lead it.
