UK businesses are investing heavily in artificial intelligence, automation and new ways of working. Yet technology alone does not guarantee better performance. When the incentives surrounding that investment are poorly designed, it can simply lead to organisations doing the wrong things faster.
Consider three familiar management decisions. Each is intended to improve performance. In each case, however, the measure encourages a response that undermines the wider objective:
- A manufacturer rewards teams for increasing output. Production rises, but so do defects and maintenance problems
- A customer service business introduces targets for shorter calls. Employees end conversations more quickly, but customers have to call back
- A procurement team is praised for delivering immediate savings. It negotiates a lower price but leaves the business reliant on a less resilient supplier.
In each case, the organisation achieved exactly what it chose to measure. People responded rationally to the incentives they faced. The problem was that the measure captured only one dimension of performance: a proxy for the outcome the business truly cared about.
The growing adoption of AI makes these trade-offs more important, not less. Technology amplifies whatever incentives already exist inside an organisation, allowing good systems to become more effective, and flawed ones to become more damaging.
This matters because UK productivity growth remains weak compared with the trends seen before the 2008 financial crisis. At the same time, the CBI’s June 2026 Economic Forecast also expected business investment to contract amid high costs and elevated uncertainty.
For firms under pressure to deliver more from limited resources, better targets, stronger incentives and closer monitoring may appear to offer an answer. But unless the underlying incentives are well designed, they can create new costs and risks:
When the metric becomes the job
Most leaders cannot directly observe everything that contributes to good performance.
They can see sales figures, production volumes, project completion dates and customer response times. It is harder to measure the quality of a relationship, the risks avoided by an experienced employee or the long-term consequences of a decision.
Economists describe this kind of problem as moral hazard: a situation in which one party cannot fully observe the actions of another, making it harder to align individual incentives with organisational goals.
Organisations respond with targets, bonuses, contracts and monitoring. These mechanisms are necessary, but they are inevitably incomplete.
A salesperson must generate revenue while protecting margins and customer relationships. A factory must increase output without compromising quality or safety. An executive must deliver current performance while making investments that may not pay off for several years.
Once one outcome is rewarded more heavily than the others, people have a reason to focus their effort there.
Leaders should therefore ask not only, “What result do we want?” but also, “What will people actually do to produce the number we have chosen?”
Stronger incentives can create stronger distortions
When performance disappoints, the instinct is often to increase the incentive.
The target becomes more demanding. The bonus becomes larger. The penalty becomes more severe. Monitoring becomes more detailed.
But strengthening an imperfect incentive can also strengthen the unintended behaviour it produces.
A bonus tied too closely to short-term profit may encourage managers to postpone maintenance or investment. A service target focused on speed may reduce the quality of customer interactions. Aggressive supplier penalties may make contractors less willing to disclose emerging problems before they become serious.
This does not mean businesses should abandon targets or performance-related rewards. It means leaders need to consider the full range of behaviour a system makes attractive.
One useful test is to imagine how a capable, rational employee or supplier could meet the target at the lowest possible cost to themselves.
Would doing so also create value for the business?
AI can scale good decisions and bad ones
Artificial intelligence can help organisations close genuine information gaps. It can reveal inefficiencies, improve forecasting and make complex operations easier to monitor.
The potential gains are already becoming visible. Government research found that 56% of businesses using AI reported an increase in employee productivity. However, 77% had not yet seen any change in revenue. [1]
That gap matters. Completing work more quickly is not necessarily the same as creating more commercial value.
An organisation that automates a poorly designed process may simply carry it out faster and at greater scale. A dashboard can make activity more visible without showing whether that activity is useful.
People may also adapt their behaviour to satisfy the system. Once employees know which activities are monitored, those activities naturally receive greater attention.
The leadership challenge is not simply to gather more data. It is to decide which information should influence decisions, what remains outside the data and where human judgement must still play a role.
Five questions for business leaders
Before introducing a new target, reward or monitoring system, leaders should ask:
- What behaviour will someone need to demonstrate to improve this measure?
- What important work or outcome is absent from it?
- Could the target be met without creating genuine value?
- Does it encourage long-term performance or the fastest visible result?
- What might people stop doing in order to succeed?
No measure will capture every aspect of performance. But these questions can help businesses identify distortions before they become embedded.
Applying economic thinking to business
These ideas are explored by Professor Ludvig Sinander in Oxford Elevate’s Economics for Business Leaders programme.
Drawing on game theory, moral hazard and incentive design, Ludvig explains how organisations can anticipate the responses of employees, competitors, customers and commercial partners before introducing a new strategy or system.
Participants gain practical frameworks to challenge assumptions, test decisions and better anticipate how people are likely to respond.
The UK’s productivity challenge requires investment, skills and technological adoption. But the return on those investments will depend on the systems surrounding them.
Technology changes what organisations are capable of doing. Incentives determine what they actually do.
The businesses that turn investment into lasting productivity gains will be those that design incentives so individual decisions consistently create long-term value.
About the contributor
This article draws on a lecture delivered by Ludvig Sinander as part of Oxford Elevate’s Economics for Business Leaders programme.
Ludvig is an award-winning Associate Professor of Economics at the University of Oxford and a Professorial Fellow at Nuffield College. His research focuses on economic theory and the design of systems that motivate people to act in desired ways, known as incentive or mechanism design.
Economics for Business Leaders is Oxford Elevate’s flagship executive education programme for senior decision-makers. Running for over 74 years, and taught by University of Oxford economists, it helps leaders apply economic thinking to real business challenges and make more confident decisions in complex and uncertain environments.
References
[1] Department for Science, Innovation and Technology, AI Adoption Research, 13 February 2026.