In 2017 the United States Bureau of Labor Statistics (BLS) published estimates from a set of questions on electronically mediated work added to the Contingent Worker Supplement. The questions asked whether, in the previous week, a person had obtained short jobs or tasks through websites or mobile applications that connected them with customers and facilitated payment. The resulting estimates were widely discussed as an early official reading of online platform labour. They were also criticised for how narrowly the concept was drawn and for how respondents interpreted the wording.
What the BLS counted
Electronically mediated work, in the BLS formulation used for those estimates, required an intermediary website or app that both matched the worker to customers and handled payment. Work arranged directly with a client by email or phone, without that intermediary payment role, fell outside the definition even if the work was done online. The BLS reported that a small share of employed people met the definition in the reference week studied. Michael Horrigan and colleagues at the BLS documented the concept, the question text and the published rates in Monthly Labor Review articles and related releases.
Why the estimates were contested
Subsequent commentary argued that the definition excluded common arrangements that the public debate treated as “gig” or platform work, and that a single reference week understates intermittent activity. Researchers including Katharine Abraham and others working on survey measurement of nontraditional employment have noted that infrequent platform work can be missed when the question window is short and when respondents do not recognise the activity as meeting the electronic-intermediation criteria. The BLS itself described the measure as experimental and subject to revision as concepts and questions evolve.
A further issue is classification as employed. Someone whose only recent platform activity fell outside the survey’s employment definitions, or who did not report it as work in the main Current Population Survey questions, would not appear in the electronically mediated counts even if a broader annual measure would have recorded the activity. Survey error and conceptual narrowness therefore compound.
Later corrections and related series
Follow-up discussion in the labour-statistics community treated the 2017 figures as a starting point rather than a settled stock measure of the platform workforce. Related research on online labour demand — for example work that scrapes vacancy or task postings from platforms — answers a different question again: it observes demand-side listings, not household reports of hours or pay. Those strands should not be averaged with the BLS electronically mediated estimates without restating units.
What this summary does not claim
This note does not assert that electronically mediated work is large or small relative to any other category of employment. It records that the BLS published a specific, tightly defined measure; that the published share was low relative to many informal claims about platform work; and that methodologists have argued the measure understates some forms of online earning activity that fall outside its wording. Readers consulting the original Monthly Labor Review articles will find the question text and the caveats attached by the BLS.
Spare Beam does not treat the BLS series as advice about any form of work. The series is summarised here as an example of how official statistics operationalise online intermediation — and of how that operationalisation has been debated in published research.
Sources named in the text · September 2026