Skip to content
Leap TrendsBeta

NLEN

What do you think of this page?

Leap Trends is in beta. Your response goes to Leap and helps us improve the pages. You can respond anonymously; the page you are on is sent along automatically. Privacy statement

How useful is this page for your work?
1 not useful5 very useful

↑ ↓ to choose · enter to openUpdated 5 November 2026 · by Leap Company

Leap Trends · Technology · AI at work

Condition

Technology

Also touches on the condition Culture · How we choose the condition

Briefing

Algorithmic management in warehouses, delivery and retail

Topics
  • Algorithmic management
  • Work stress
  • Regulation

Algorithmic management is spreading from platform work to warehouses, transport and shops. A Swedish study of 978 drivers and warehouse workers links heavy algorithmic management to twice the rate of mental health complaints. Brussels and The Hague are drafting rules.

Oldest publication Updated

Compiled with AI from public sources, with the source for every figure. Key figures labelled Checked have been compared with the source by a person. How that works

Key figures

What the sources say

In algorithmic management, software takes over tasks that used to belong to a manager: allocating work, measuring the pace, drawing up rosters and assessing performance. Researchers from TNO and three other European occupational safety institutes write in the first 2026 issue of the Scandinavian Journal of Work, Environment & Health that such systems are spreading quickly from platform work to logistics, retail and care. Their review of 39 studies from 2022 to 2024 shows that the systems often increase workload while reducing autonomy and contact with colleagues, especially when the way they work is opaque or tasks are tightly timed and tracked. According to the authors, giving employees a say and explaining how the system works softens those effects.

A Swedish study, published in November 2025 in the International Archives of Occupational and Environmental Health, measures this in logistics. Between February and July 2024 the researchers surveyed 978 workers online: 592 drivers and 378 warehouse workers, recruited through social media and union newsletters. They measured how often tasks, shifts and instructions were assigned automatically, whether work was tracked digitally and whether performance was assessed automatically. About 9 per cent were subject to heavy algorithmic management. In that group mental health complaints were 2.1 times as common as among those with little or none, workplace accidents with sick leave 1.9 times, headaches 1.7 times and muscle and joint complaints 1.5 times. The associations were stronger for drivers than in the warehouse. This is a snapshot, so cause and effect are not established.

Tiago Alexandre Vieira and Pedro Mendonça of Heriot-Watt University look beyond a single sector. In an exploratory article from late October 2026 they use a representative European survey of workers to examine how automated allocation of work, among other things, relates to safety and health at work. They read algorithmic management through Labour Process Theory, as part of the struggle over control of work, and criticise the debate for treating the technology separately from the decision-making room that management keeps for itself. The title speaks of a contested relation and of non-deterministic insights. What that means for the outcomes is not stated in the abstract, which gives no figures and does not name the survey either.

In shops it is mainly about the roster. For the Talking Shop 2026 report, Retail Week and Rotageek, which sells rostering software, surveyed 500 UK shop workers. Of these, 71 per cent call the rostering in their own store reactive, with last-minute changes and too many or too few people on the shop floor. At the same time, 76 per cent say they would trust an AI system with the roster in some form.

Regulation is moving too. On 17 December 2025 the European Parliament adopted a resolution asking the Commission for a legislative proposal on algorithmic management at work. Parliament wants continuous human oversight of decisions such systems take or prepare, written explanations for employees and their representatives, consultation when introducing systems that affect pay, appraisal, task allocation or working time, and a ban on processing emotional and biometric data. For platform work there is already a European directive from 2024. On 29 June 2026 the Dutch Ministry of Social Affairs and Employment put the draft Platform Work Act out for consultation. Anyone who works as a driver or delivers meals via a platform will then be able to require the platform to explain the algorithm’s decisions, must be told when the way it works changes, and dismissal or suspension may only be decided by a human.

How solid is this? Researchers in several countries find the same direction, from a network of European occupational safety institutes to a study in Swedish logistics, and both Brussels and The Hague are working on rules. The Heriot-Watt article adds a representative European sample, although its results cannot yet be followed in figures. Beyond that, the evidence from warehouses and transport consists mainly of snapshots from northern Europe, Dutch measurements are missing and the retail figures come from a software vendor. So the picture is not yet settled.

For the leadership team

Letting software allocate tasks, measure the pace and draw up rosters hands part of managing to a system. Does the leadership team know which decisions on targets, rosters and appraisal are now prepared or taken by software? And can anyone explain those decisions to the employee and the works council, as the Dutch Platform Work Act will require of platforms?

Sources

  1. Rijksoverheid (ministerie van SZW)Gelijker speelveld en betere positie platformwerker door nieuwe wet · policy · Transport and logisticsBelangrijke beslissingen, zoals ontslag of schorsing, mogen dan alleen nog door een mens worden genomen.
  2. Rotageek and Retail Week · Amy RosomanWhy AI adoption stalls on the shopfloor, and how retail leaders can fix it (Talking Shop 2026) · survey · Retail and consumer
  3. Scandinavian Journal of Work, Environment & Health (PEROSH network, with TNO) · Mairi Bowdler, Heidi Lahti, Marie Jelenko, Giuliana Buresti and Teppo ValtonenAlgorithmic management and psychosocial risks at work: An emerging occupational safety and health challenge · opinionThe impact of ALMA on workers ultimately depends on managerial choices and the broader organizational context.
  4. European ParliamentDigitalisation, artificial intelligence and algorithmic management in the workplace: shaping the future of work (2025/2080(INL)) · policyThe proposal should ensure continuous, meaningful and effective human oversight at all times over all decisions taken or supported by algorithmic management systems.
  5. International Archives of Occupational and Environmental Health · Karin Hennum Nilsson, Theo Bodin, Pille Strauss et al.Algorithmic management is associated with psychological distress, musculoskeletal pain, and occupational accidents: a cross-sectional study in logistics · research · Transport and logisticsThese findings highlight AM as a potential occupational health hazard, particularly when involving high levels of automated oversight and direction.

We summarise in our own words and quote no more than one sentence per source. Read the original: there is more in it. How we use sources.

Academic research

  1. Heriot-Watt University · Tiago Alexandre VIEIRA and Pedro MendonçaAlgorithmic management and occupational health and safety: Non-deterministic insights into a contested relation · researchDebates often isolate algorithmic management from the structural dynamics of managerial prerogative, overlooking how they relate to OHS risks.
  2. Széchenyi István University · Zelles et al.Beyond Platform Work: Algorithmic Management in Platform Labor and Traditional Organizations—The Hungarian Experience · research
  3. University of Duisburg-Essen · Müller et al.The association between core functions of algorithmic management and work-related health problems: indications for effects beyond well-known psychosocial stress · research
  4. Educons University · Milojević et al.Algorithmic Management and Employee Outcomes: Dual Effects of Control and Capability on Autonomy and Career Security · research
  5. Kozminski University · Miszczyński et al.The algorithm’s middlemen: resistance, compliance and managerial contradictions in Amazon’s warehouses in Poland · research · Transport and logistics
  6. Karolinska Institutet · Lind et al.(Un)healthy algorithms? Trade union, employer, and government action for safety and health in algorithmically managed work · research

Academic articles on the same phenomenon, no more than a year old at the last search, from universities in Europe and North America, found through the open database OpenAlex. These articles count towards the publications per month; the text and key figures above are based on the sources. How we choose the articles.

How this briefing has been updated

  1. Added an exploratory article from Heriot-Watt University that examines the link between algorithmic management and occupational safety using a representative European survey of workers; the abstract does not yet give any figures. (AI)

  2. Title now describes the topics; previously “In warehouses, delivery and retail, an algorithm increasingly divides up the work” (AI)

  3. Added 1 article from universities (via OpenAlex) after a search round for new topics and sources. (AI)

  4. Added 4 articles from universities in Europe and North America (via OpenAlex). (AI)

  5. First draft based on a Swedish study in logistics, a review by the PEROSH network, a UK survey of shop workers and legislation in Brussels and The Hague. Dutch measurements in warehouses are missing. (AI)

More briefings

Technology

AI in software development and the trouble with measuring productivity

  • AI in software development
  • Developer productivity
  • Coding agents

The share of developers willing to leave application logic to AI rose from 18 to 50 per cent in a year. Measuring the return no longer works: METR cannot find enough developers still willing to work without AI, and 66 per cent feel metrics miss their contribution.

Publications per month for AI in software development and the trouble with measuring productivityDecember 2025: 0 publicationsJanuary 2026: 1 publicationFebruary 2026: 2 publications2March 2026: 1 publicationApril 2026: 1 publicationMay 2026: 0 publicationsJune 2026: 2 publicationsJuly 2026: 0 publicationsAugust 2026: 0 publicationsSeptember 2026: 0 publicationsOctober 2026: 2 publicationsNovember 2026: 0 publications
Publications per month for AI in software development and the trouble with measuring productivity
MonthPublications
December 20250
January 20261
February 20262
March 20261
April 20261
May 20260
June 20262
July 20260
August 20260
September 20260
October 20262
November 20260

updated · 11 publications

Technology

Shadow AI: employees' own AI tools and the policy lagging behind

  • Shadow AI
  • AI policy
  • Information security

Many employees use their own AI tools alongside those provided, often paying for them themselves, because they work better or faster. Managers greatly overestimate how much they can see. Both the risk and the opportunity stay invisible as long as nobody asks.

Publications per month for Shadow AI: employees' own AI tools and the policy lagging behindDecember 2025: 0 publicationsJanuary 2026: 0 publicationsFebruary 2026: 0 publicationsMarch 2026: 0 publicationsApril 2026: 3 publications3May 2026: 2 publicationsJune 2026: 0 publicationsJuly 2026: 2 publicationsAugust 2026: 0 publicationsSeptember 2026: 2 publicationsOctober 2026: 0 publicationsNovember 2026: 0 publications
Publications per month for Shadow AI: employees' own AI tools and the policy lagging behind
MonthPublications
December 20250
January 20260
February 20260
March 20260
April 20263
May 20262
June 20260
July 20262
August 20260
September 20262
October 20260
November 20260

updated · 10 publications

Technology

Agent-washing and the cost and value of agent projects

  • Agent-washing
  • Value of agent projects
  • Managing agents

The number of AI agents is growing fast, but Gartner expects over four in ten agent projects to be cancelled by the end of 2027 because of rising costs and unclear value. Vendors, meanwhile, are sticking the agent label on existing products. The value goes to companies that manage agents as a single programme.

Publications per month for Agent-washing and the cost and value of agent projectsDecember 2025: 0 publicationsJanuary 2026: 0 publicationsFebruary 2026: 0 publicationsMarch 2026: 0 publicationsApril 2026: 1 publicationMay 2026: 1 publicationJune 2026: 1 publicationJuly 2026: 0 publicationsAugust 2026: 1 publicationSeptember 2026: 3 publications3October 2026: 1 publicationNovember 2026: 0 publications
Publications per month for Agent-washing and the cost and value of agent projects
MonthPublications
December 20250
January 20260
February 20260
March 20260
April 20261
May 20261
June 20261
July 20260
August 20261
September 20263
October 20261
November 20260

updated · 9 publications