The human price of dynamic pay
Platform operators are now using dynamic pay to boost their profits and exert greater power over workers.
This means that different workers may be offered different rates for the same job, determined by an algorithm whose operation is a mystery to those subject to it.
Workers, for their part, can no longer accurately predict what they will earn when they head out for a shift.
The prospect of collective bargaining between unions and employers, still limited in the UK platform economy, is undermined because there is not a wage rate to negotiate over.
The case studies in this report shows the human side of this development including the effect of workers’ stress levels and the toll it takes on family life.
They make a compelling case for swift government action to put the legal protections in place to ensure that platform work is dignified work.
This report therefore argues for action in three linked areas: ending dynamic pay-setting, urgent reform of employment status, and collective data rights that enable effective transparency and bargaining.
Dynamic pay now sits at the centre of platform work in the UK, yet it remains one of the least understood transformations in modern labour markets. In this report, ‘dynamic pay’ refers to algorithmically-determined, variable and potentially personalised worker compensation for a task. This can include variable commission or take rates and other adjustments that make workers’ pay unpredictable at the point of accepting work.
Many platform operators now use algorithms to set an individual price for each job that can vary from person to person and from job to job. This is presented as neutral or even beneficial: matching supply and demand in real time and rewarding workers for being in the right place at the right moment. This approach might be familiar, if not always liked, by many people in areas like concert or airline ticket pricing. But when applied to pay this represents a profound transformation of how wages are calculated, distributed, and justified. And workers are losing out.
Dynamic pay replaces fixed rates or transparent tariffs with opaque, constantly shifting pricing mechanisms. Under this system, two workers performing the same task may be paid entirely different amounts; a worker performing identical tasks across two days may receive vastly different pay; and individuals have almost no ability to anticipate income, plan financially, or verify the accuracy or fairness of their compensation. It is sometimes not even clear to workers how much they will be paid at the point they must decide whether to accept or reject an offer of work.
For workers, pay becomes decoupled from time, skill, or effort. Instead, it becomes a speculative outcome of an algorithmic process that remains largely invisible to those whose livelihoods depend on it. Workers describe themselves as “gambling,” “leaving it to fate,” or “waiting for the jackpot,” because pay feels like the outcome of chance rather than work.
Dynamic pay functions as a wage-setting regime that structurally redistributes risk onto workers and weakens their capacity to anticipate earnings, contest pay outcomes, or negotiate collectively, regardless of whether any individual pricing decision is designed to achieve that outcome.
As the case studies in this project show, dynamic pay incentivises workers to accept less favourable jobs, stay online longer, and work more unpredictable hours, while expanding unpaid waiting time and producing high levels of income volatility. One study found that riders and drivers report spending an average of ten hours a week waiting for jobs to come through on the app – so logged on and working but not making any money.[1]
The spread of dynamic pay has not occurred in a vacuum. Ambiguities over employment status allow platforms to say that those who work for them are workers with limited employment rights or, in some cases, self-employed workers with effectively no rights at all. This is despite exerting managerial control through algorithms and frequent requirements regarded the use of branded equipment and meeting company-determined standards.
This is coupled with the deliberate oversupply of labour that gives platforms the power to introduce extreme variability in pay. With tens of thousands of workers competing for a limited pool of jobs, a platform can offer rock-bottom wages and undermine pay certainty, confident that someone will always accept the work.
This structural oversupply interacts with other enabling conditions. Under the prevailing platform model, in areas like ride hail and food delivery, workers are not paid for waiting time: meaning they bear the risk if business is slower than expected. Meanwhile, state enforcement of employment and data rights is weak. Trade unions, whose reps provide a first line of defence against exploitation, have gained a foothold with national agreements at some operators. But the atomised nature of the workforce presents a significant barrier to building the collective strength needed to provide a strong countervailing force against the financial and technological power of platform employers backed by the UK’s restrictive union laws.
At the same time, platform operators are trying to shift from expansion to profitability making it a priority to squeeze more value from workers.[2]
In this environment, dynamic pay operates by continually probing what rates workers will accept and shaping the distribution of work around those responses. The result is a wage-setting regime able to introduce wide variability and downward pressure on earnings because the labour market has been engineered to absorb it. Together, these conditions have created a system in which workers’ incomes are determined by processes they cannot see, influence or meaningfully challenge, either individually or collectively.
The most comprehensive evidence on dynamic pay in the UK comes from the 2025 study conducted by University of Oxford and Worker Info Exchange (Not Even Nice Work If You Can Get It)[3], which analysed 1.5 million trips by 258 Uber drivers between 2016 and 2024. The findings support what workers have described: dynamic pay is not simply more sophisticated pricing, it is a new wage-setting regime that reduces pay, increases inequality, and makes stable income impossible.
Central to the study is the finding that after Uber introduced dynamic pricing in 2023:
Platform take rates increased substantially. Uber had long presented a nominal 25 per cent commission, but dynamic pricing transforms this into a variable and personalised take. The study found that drivers now frequently retain only 50–60 per cent of the fare, with Uber taking over half on many trips. The median driver retained just 71 per cent, and only 46 per cent of drivers retained 75 per cent or more. Uber’s surplus per hour increased by 38 per cent following the introduction of dynamic pay.
Drivers’ real pay declined and became more volatile. Taking into account all the worker’s time logged onto the app when they were available for work[4], average pay per hour fell in real terms after dynamic pricing. Even under Uber’s narrower definition (“engaged time” only), real hourly pay declined.
Unpaid working time increased significantly. Drivers now spend more time waiting for work than performing trips, meaning much of their working day generates no income.
Predictability collapsed. Using machine learning models, researchers demonstrated that earnings for similar trips could no longer be predicted based on past patterns. Pay became a function of algorithmic behaviour that workers could not anticipate or influence.
Inequality between workers increased. The study found a widening gap between “winners” and “losers,” even among workers doing comparable jobs. Post-dynamic pricing, 93 out of 114 drivers earned less, while only 21 earned more. Drivers doing the same work increasingly receive different pay due to what appears to be personalised algorithmic decisions.
These findings confirm that dynamic pay is not simply confusing. It systematically redistributes value from workers to platforms and erodes the preconditions of fair wage-setting and the principle of same job, same pay.
This is occurring at a time when the government is pursuing its Plan to Make Work Pay aimed at overhauling the labour market. Early reforms, which have recently received parliamentary approval[5], seek to deal with exploitation at the sharp end of the labour market, for instance by giving variable hours workers a route onto secure contracts and making it harder for employers to dismiss employees without cause. It also paves the way for a stronger collective voice in the economy by providing unions with a right to access workplaces (including digital access) to talk to workers; dismantling some of the barriers to union recognition; and dissolving some of the anti-worker legislation of the last 10 years.
What this report shows is that more is needed to tackle the novel form of exploitation taking place in the platform economy.
We propose three key sets of measures:
Ending the use of dynamic pay. The government, in its Plan to Make Work Pay, is committed to jobs that provide workers with fair pay and security. Dynamic pay is not the logical next stage in development of the platform economy. Rather as the evidence shows, dynamic pay increases the insecurity of worker’s wages, and undermines the principle of ‘same pay for same job’. It is not the logical next stage in development of the platform economy. Under the guise of ‘innovation’ it is little more than a throwback to a bygone era where employers did their utmost to disguise how wage-setting worked.[6]
Urgent employment status reform. The government, in its Plan to Make Work Pay, is already committed to reform of employment status, including moving towards a single status of worker away from the current division between workers and employees. This must be a priority. Employment status is the fundamental building block towards secure work, fair pay and collective bargaining, which are all government aims. Without a sound employment status regime, exploitation of workers will only get worse. A particular focus should be put on tackling bogus self-employment.
Collective data rights for trade unions. The TUC, working with a cross-party group of stakeholders has set out a model for how this might work in a model AI Bill. Power in the workplace is already tilted towards employers and operators. Algorithmic dynamic pay-setting skews it further. Only by allowing the pooling of worker data and placing transparency obligations on employers and operators can some of this imbalance start to be addressed.
The following case studies examine how dynamic pay shapes the everyday realities of platform economy workers. They capture the uncertainty and stress created when pay fluctuates according to opaque algorithms, highlighting how this affects workers’ ability to plan, earn a stable income, and maintain their wellbeing.
Each account illustrates the broader human cost of a system that transfers risk from platforms to individuals, revealing how insecurity and employer power extend into workers’ personal lives.
[1] Wood, A. et al (June 2025). “Beyond the ‘Gig Economy’: Towards Variable Experiences of Job Quality in Platform Work”, Work, Employment and Society Volume 39, Issue 5 https://journals.sagepub.com/doi/10.1177/09500170251336947
[2] Stacey, S. (8 October 2024). “Online gig platforms focus on profits as workers return to office”, Financial Times www.ft.com/content/6189ba99-df17-4f80-aca2-4214a482bb98
[3] Binns, R. et al. (2025), “Not Even Nice Work If You Can Get It; A Longitudinal Study of Uber’s Algorithmic Pay and Pricing,” arXiv preprint arXiv:2506.15278 https://doi.org/10.48550/arXiv.2506.15278
[4] This was the employment tribunal’s definition of working time in Uber BV v Aslam www.supremecourt.uk/cases/uksc-2019-0029. In Ashfar and others v Addison Lee, a judgment handed down in January 2025, the tribunal ruled that all passenger drivers, courier drivers and executive drivers are working for the company during the times they are logged onto its app or mobile device https://oldsquare.co.uk/wp-content/uploads/2025/01/Afshar-Ors-v-Addison….
[5] Pickard, J. and Strauss, D. (16 December 2025). “UK government’s flagship workers’ rights legislation clears final hurdle”, Financial Times www.ft.com/content/60f5cdde-7b60-4d88-9033-32812ead2da0
[6] Dubal, V (21 August 2025). How artificial intelligence uncouples hard work from fair wages through ‘surveillance pay’ practices—and how to fix it. Washington Center for Equitable Growth https://equitablegrowth.org/how-artificial-intelligence-uncouples-hard-…
Stay Updated
Want to hear about our latest news and blogs?
Sign up now to get it straight to your inbox