The 114th session of the International Labour Conference, held in Geneva from 1 to 12 June 2026, marks one of the most consequential moments in international labour standard-setting since the adoption of Convention 189 on domestic workers in 2011. At its centre is a deceptively simple question: when a worker earns a living through a digital platform, do they have rights?
The answer, after two years of tripartite negotiation, is yes — and the implications for how NSOs measure labour markets are immediate.
What the Convention Does
The new ILO instrument on decent work in the platform economy abandons the classical employee/self-employed dichotomy for the purposes of worker protection. Instead of asking whether a platform worker has a contract of employment, it asks whether they perform work through a digital labour platform — and if so, it attaches a broad floor of rights: freedom of association, occupational safety and health protection, social security coverage, data protection, and access to remedy.
This is not a marginal adjustment. It is a structural reclassification. For labour statisticians, it immediately raises a question that no current LFS questionnaire is designed to answer cleanly: how many workers in your country work through a platform, as opposed to working for a platform as an employee, or with a platform as a business?
The Measurement Gap
Most national LFS instruments can identify the self-employed, distinguish employees from own-account workers, and capture multiple job-holding. Very few can identify platform-intermediated work as a distinct category of labour market attachment. The ILO's own research consistently finds that platform workers are scattered across conventional employment status categories — some classified as employees, many as own-account workers, some not captured at all if the work is irregular or secondary.
The first EU-wide measurement of platform work, conducted in 2022 across 17 Member States and EFTA countries, puts the scale of the phenomenon in concrete terms. Approximately 3% of the population aged 15–64 worked through a digital platform for at least one hour in the preceding 12 months. The figure is modest in aggregate but the distribution tells a more complicated story. When the same workers are examined by labour status, more than one in eight falls outside what the core LFS would classify as employed, meaning they were either unemployed or outside the labour force at the time of the survey. They worked. They were paid. But the existing statistical machinery did not classify them as employed.
This is not a rounding error. It is a structural measurement gap, one that the ILC 114 Draft Convention addresses directly. Article 2(1)(b) of the draft instrument extends its scope to all digital platform workers regardless of their employment relationship or contractual arrangements, explicitly covering those in the informal economy. For this group, the question of employment classification does not arise in the conventional sense, there is no employment relationship to classify. The Convention reaches them regardless. For National Statistical Offices, this creates a concrete design challenge. An LFS module that screens only employed persons or only the self-employed will structurally miss part of the population the Convention is intended to protect. The placement decision is not merely a technical choice about questionnaire flow. It is a decision about who gets counted.
The new Convention creates a legal category without a corresponding statistical one. That gap will need to close. NSOs that have already piloted supplementary platform work modules — as several European NSOs have done in coordination with Eurostat — will have a head start. Those that have not will face pressure from ministries of labour, social security agencies, and the ILO itself to produce evidence on a phenomenon that is now internationally regulated but remains statistically invisible in most countries.
"The Convention creates a legal category without a corresponding statistical one. That gap will need to close."
Where AI Enters
The ILC's headline message from ILO Director-General Gilbert Houngbo was precise: "The future of work will not be determined by technology alone." That statement is both reassurance and provocation. Technology — including AI — is reshaping work faster than statistics can track it. The ILC 114 agenda included labour standards for platform workers alongside questions about how AI-driven management systems, algorithmic task allocation, and automated performance monitoring are changing the nature of work itself.
For NSOs, AI is simultaneously the subject being measured and the tool available to measure it. AI-assisted coding of occupation and industry from free-text responses can improve coverage of non-standard workers. Machine learning applied to administrative data — tax records, social insurance registrations, platform transaction logs — can identify platform workers who are invisible in household surveys. Natural language processing can flag responses where the described work pattern is inconsistent with the reported employment status.
None of this replaces the survey. But the combination of a new international legal standard, a growing population of workers it covers, and maturing AI tools for statistical production creates an alignment that NSOs should act on deliberately — not reactively.
What NSOs Should Do Now
Three practical steps follow directly from what was agreed at ILC 114.
1. Review your LFS questionnaire for platform work coverage
The minimum requirement is a filter question that identifies workers whose tasks are allocated or mediated by a digital platform, regardless of their employment contract. This does not require a full questionnaire redesign — it requires a targeted module, tested for cognitive validity against the ILO's own platform economy definitional framework.
In most countries, tax authorities and social security agencies hold records that, with appropriate legal gateways, could illuminate the scale and composition of platform work more precisely than any household survey conducted annually. Mapping those sources now — before reporting frameworks are mandated — puts the NSO in a position of strength rather than reaction.
3. Begin capacity building now
The Convention, once ratified, will require national reporting. Those reports will need statistical evidence. The NSO that is not at the table when reporting frameworks are designed will find itself producing statistics to fit a framework built without it.
The Choice
The ILO Director-General's framing — a moment of choice — captures something real. The decision is not whether AI will affect labour markets. It already has. The decision is whether international statistical infrastructure will keep pace, or whether the gap between what is regulated and what is measured will widen until it becomes a governance failure.
National Statistical Offices are not bystanders to this process. They are its foundation.