The Manager Outside the Firm
Researched by an agentic pipeline · reviewed and gated by the author
If the enterprise can control access to work but cannot explain the decision, it has created managerial power without managerial accountability.
The manager outside the firm
An employment contract once offered a reasonably clear map of managerial power. The organisation hired the worker, a manager allocated work, performance was reviewed and the employer carried the corresponding obligations.
That map no longer fits the territory.
A courier can be formally independent while software sets the price, offers the task, measures acceptance, ranks performance and suspends access to future income. A contractor may work through an intermediary while the buying enterprise’s system determines schedules and quality thresholds. An employee can be managed by an algorithm whose owner sits in technology, whose data comes from operations and whose decision no individual manager can fully explain.
The legal categories remain important. But the practical boundary of management is moving.
On 12 June 2026, the International Labour Conference adopted Convention No. 193, the first binding international labour standard designed for the platform economy. It addresses pay, safety, social protection, termination and algorithmic management across formal classification boundaries. Platforms must disclose how automated systems affect work and provide human involvement for consequential decisions. [S1] [S4]
The Convention is not immediately binding worldwide. It requires ratification and domestic implementation, and employment status still affects how protections apply. Yet its direction converges with the European Union’s Platform Work Directive, which already establishes transparency, data and human-oversight rights for persons performing platform work and must be transposed by member states by December 2026. [S3]
Together they point to a deeper institutional shift: workforce accountability is beginning to follow the exercise of managerial power, not only the formal perimeter of employment.
Management without a manager
Algorithmic management is not simply the use of AI in HR. It is software performing or shaping functions traditionally associated with management: instructing, monitoring, evaluating, rewarding and disciplining people.
The OECD surveyed more than 6,000 firms across France, Germany, Italy, Japan, Spain and the United States. It found such tools already common in most countries studied. Managers reported benefits in decision quality and job satisfaction, but also concerns about unclear accountability, opaque logic and worker health. [S6]
The organisational problem is easy to miss because each component has a legitimate owner. Procurement contracts with a platform. Operations defines service levels. Data teams build monitoring. HR maintains employment policy. Legal reviews terms. No one sees the combined system as management.
Consider an illustrative retailer using three labour channels in the same fulfilment centre.
Employees receive shifts through an internal scheduling system. Agency workers are supplied by a staffing firm but ranked using the retailer’s productivity data. Independent couriers accept delivery offers through an external platform whose price and deactivation logic responds to retailer demand.
The contracts differ. The managerial functions overlap. Software determines who sees work, when it appears, what it pays, how performance is judged and whether future access continues. If governance follows only the contract label, the enterprise can apply three different standards to the same exercise of power.
The function map
Leaders need a second view of the workforce—not a headcount map, but a management-function map.
| Function | Governing question |
|---|---|
| Access | Who decides whether a person can obtain work? |
| Allocation | What system assigns tasks, shifts or customers? |
| Pay | Who determines price, incentive, deduction and expense treatment? |
| Monitoring | What data observes behaviour, location and output? |
| Evaluation | How are scores, rankings and quality judgments produced? |
| Discipline | What can reduce work, suspend access or end the relationship? |
| Appeal | Can a person obtain an explanation and meaningful human review? |
For each function, identify the entity with practical control, the system making or supporting the decision, the data used, the individual accountable and the route for challenge. Apply the map to employees, contractors, platform workers, franchisees and supplier workforces where enterprise systems materially direct performance.
This does not imply that every commercial relationship becomes employment. It reveals where the organisation exercises management-like power and therefore faces operational, legal and legitimacy risk even when formal responsibility is contested.
If the enterprise can control access to work but cannot explain the decision, it has created managerial power without managerial accountability.
Rights following control
Platform work made the mismatch visible first.
The EU directive recognises that automated monitoring and decision systems can shape working conditions regardless of whether the platform regards itself as an employer. It limits certain data processing, requires information and impact assessment, and reserves suspension or termination decisions for human involvement. [S3] Academic analysis describes this as a synthesis: traditional employment protections remain linked to status, while some digital rights extend to all persons performing platform work. [S8]
Convention No. 193 creates a global floor with similar logic. It does not impose one classification model. It requires reality, not only contractual description, to determine status, and establishes protections that reach workers in different formal categories. The adoption vote—406 in favour, eight against and 36 abstentions—shows substantial support and real division. The United States opposed a prescriptive convention in a fast-evolving economy, warning that rigidity could hinder innovation. [S4]
That objection matters. A rule designed for ride-hailing can be poorly fitted to a specialist marketplace, a cooperative or a genuinely independent business-to-business platform. Human review can become a rubber stamp. Detailed disclosure can allow gaming or expose sensitive fraud controls. National implementation will vary.
The right response is proportional governance, not denial that management is taking place.
High-consequence decisions—pay, access, safety, suspension and termination—need traceable rules, accountable humans and an appeal capable of changing the outcome. Lower-consequence recommendations may need monitoring and sampling rather than case-by-case approval. The standard should follow consequence and power, not an assumption that every algorithm is equally risky.
The strongest limit
Procedural fairness is not economic justice.
Research from the Fairwork project argues that transparency, explanation and data rights do not by themselves solve low pay, insecurity or weak social protection. A worker can receive a perfect explanation of an unaffordable rate. A human can review a deactivation while the underlying market power remains unchanged. [S8]
This is the strongest challenge to an enterprise governance response. Mapping algorithms and installing appeals may make decisions cleaner without changing the distribution of value or risk.
The distinction should be explicit.
Algorithmic-management controls address how power is exercised: visibility, data limits, review, consistency and contestability. Employment classification, minimum pay, benefits, bargaining and social insurance address what people are substantively entitled to receive. Neither substitutes for the other.
Convention No. 193 includes both procedural and material protections, but its force depends on ratification, national law and enforcement. The ILO itself has no direct global enforcement power. [S2] [S4] Firms should not mistake adoption for harmonisation.
From platform exception to workplace model
The argument that these rules will spread beyond platforms remains a projection.
Current legal instruments have defined scopes. Ordinary employers using workforce analytics are not automatically governed by the ILO platform convention. A contractor is not covered merely because the buying organisation uses software.
There are, however, credible signs of diffusion. The European Parliament has studied algorithmic management across logistics, healthcare, telecommunications, automotive and manufacturing, estimating exposure could rise from 42.3 to 55.5 per cent in the medium term and identifying protection gaps outside platforms. [S7] The OECD evidence shows the underlying practice is already widespread. Academic analysis calls the Platform Work Directive a potential blueprint for broader regulation of casual work and algorithmic management. [S8]
Policy diffusion often begins this way. A novel context exposes a mechanism that existing categories cannot govern. Rights are built for the visible case. Similar power asymmetries appear elsewhere. Courts, regulators, collective agreements and corporate standards adapt the principle before one universal law arrives.
That process may stall. Governments may not ratify the ILO Convention. National definitions may remain narrow. Intermediaries may fragment responsibility. A general EU workplace instrument may never emerge. The useful enterprise action does not depend on predicting the exact legal path.
It depends on recognising where consequential management is already occurring.
Govern the decision, not the box
Organisations often govern systems by administrative ownership. HR tools follow HR policy. Operational optimisation follows process controls. Supplier platforms follow procurement rules. AI models follow technology governance.
Algorithmic management crosses every box.
A credible operating model assigns one accountable owner for each consequential workforce decision, even when several systems and entities contribute. It requires documentation of the purpose, data and logic; tests for error and uneven impact; preserves logs; and provides a human reviewer with authority, time and information to reverse the outcome.
The reviewer cannot be a ceremonial endpoint. If a platform suspends a worker based on fraud risk, the human must see the evidence, understand the threshold and restore access when the signal is wrong. If an internal scheduling system systematically allocates undesirable shifts, the organisation must be able to diagnose whether the pattern reflects demand, availability data or biased optimisation.
Procurement must extend the same requirements to vendors and intermediaries. Contracting away the system should not contract away the ability to explain a decision made in the enterprise’s service.
The boundary of control
The firm of the future will not have one clean workforce perimeter. It will coordinate employees, contractors, platforms, suppliers, agents and automated systems in overlapping production networks.
Formal classification will continue to determine important rights and obligations. It will become a less complete description of who manages whom.
Convention No. 193 is a milestone, not a universal regime. The EU directive is a live legal obligation within its scope, not a general law of work. Both nevertheless establish a durable proposition: when software exercises consequential control over access, pay and livelihood, someone must disclose, oversee and answer for it.
Executives should begin with the function map. Find where the organisation allocates, monitors, evaluates and disciplines work across every contractual boundary. Identify decisions no accountable human can explain or change. Close those gaps before a regulator, worker or operational failure reveals them.
The enterprise boundary is increasingly porous. The boundary of control is not. Wherever managerial power travels, accountability will be asked to follow.
Sources
- International Labour Organization — Convention No. 193: Decent Work in the Platform Economy — 12 June 2026 — https://www.ilo.org/sites/default/files/2026-06/ILC114-CNP-D4-Final%20text-EN.pdf
- ILO NORMLEX — C193 Decent Work in the Platform Economy Convention — 2026 — https://normlex.ilo.org/dyn/nrmlx_en/f?p=NORMLEXPUB:12100:0::NO:12100:P12100_INSTRUMENT_ID:4496059:NO
- European Union — Directive (EU) 2024/2831 on improving working conditions in platform work — 23 October 2024 — https://eur-lex.europa.eu/eli/dir/2024/2831/oj/eng
- Reuters — UN labour organisation sets first global standards for gig workers — 12 June 2026 — https://www.reuters.com/business/world-at-work/un-labour-organization-adopts-convention-set-employment-standards-gig-workers-2026-06-12/
- Ogletree Deakins — ILO Adopts First Global Labor Standard for Platform Work — 12 June 2026 — https://ogletree.com/insights-resources/blog-posts/ilo-adopts-first-global-labor-standard-for-platform-work-what-u-s-companies-need-to-know/
- OECD — Algorithmic management in the workplace — 6 February 2025 — https://www.oecd.org/en/publications/algorithmic-management-in-the-workplace_287c13c4-en.html
- European Parliament — Digitalisation, artificial intelligence and algorithmic management in the workplace — 24 October 2025 — https://www.europarl.europa.eu/thinktank/en/document/EPRS_STU(2025)774670
- Industrial Law Journal — Fair Work for Platform Workers: Lessons from the EU Directive — 27 June 2025 — https://academic.oup.com/ilj/article/54/3/425/8176731