Eleven million possible job transitions call for pathways, not a forecast quota

McKinsey Global Institute’s Workforce in motion models a midpoint US scenario in which 11 million workers change occupations by 2035. Alternative assumptions produce a range of 6 million to 16 million. The report says 36 million workers could be in occupations with reduced labour demand and 41 million in occupations with increased demand, while many shifts are absorbed within the same occupation.

The method combines Bureau of Labor Statistics data, Lightcast job postings, an input-output model and assumptions about AI adoption and labour demand. McKinsey says a large language model helped synthesise public information about credentials and training requirements. The authors describe estimates as directional, not deterministic. Axios reports the headline and the difficulty of converting broad demand into practical transitions.

Keep the scenario range visible

Eleven million is not a target, probability or observed displacement count. It is a midpoint generated under assumptions. Workforce plans should show outcomes at 6 million, 11 million and 16 million transitions, plus a no-policy-change baseline. Record which investments remain useful across scenarios and which depend on one adoption path.

Keep those assumptions beside every budget decision so later evidence can update the plan.

The report estimates that roughly 25 million workers in lower-demand occupations could move to growing roles within their existing occupation, while 11 million might need cross-occupation transitions. It also identifies about 16 million openings requiring entrants from outside the occupation, including new labour-market entrants. These quantities describe modelled matching pressure, not proof that particular people can or should move.

Test the pathway, not just skill similarity

For each origin and destination role, document prerequisites, time to competence, direct cost, lost earnings, geography, schedule, licensing, accessibility, employer recognition and starting wage. Add the number of real vacancies that accept the proposed evidence. A pathway that looks close in a skills vector can still fail because training is unavailable at the right time or employers demand experience entrants cannot obtain.

The report’s LLM-assisted synthesis of public training requirements is useful for discovery but should not become the authority for eligibility. Validate requirements against current licensing bodies, providers and employers. Sample pathway recommendations for missing constraints and version every source.

The counterargument is that detailed maps will age quickly. That is true. Use them as monitored products, not static catalogues. Track vacancy volume, entry requirements, completion, placement, time to placement, wage change and retention. Retire pathways when recognition or demand disappears. Publish coverage gaps instead of forcing every displaced role into an optimistic match.

Use the Skills Intelligence Role Dictionary to standardise outcomes and capabilities on both sides of a transition, then attach local evidence. Do not infer that shared labels guarantee transfer. Prior learning, portfolio work, supervised practice and employer assessment should make the bridge testable.

The immediate decision is to fund a small set of high-volume pathways that survive the full scenario range and disclose their constraints. The 11 million headline signals possible scale; the actionable unit is a verified route from one role to another, with an employer willing to recognise the evidence.