UK workforce outlook, autumn 2026 to autumn 2027: selective hiring, persistent skills constraints and a weaker entry pipeline

Thesis: a quieter hiring market will make workforce decisions more consequential, not less

**Forecast horizon: September 2026 to September 2027, with implications for the following recruitment cycle.** Sanctuary’s central judgement is that UK employers will continue to hire selectively rather than return quickly to broad-based headcount expansion. The constraint is not simply weak demand. It is the combination of uneven demand, continued attention to operating costs, cautious replacement hiring and investment in technologies intended to raise output from existing teams.

The current data support a subdued starting point. UK vacancies fell by 6,000 to 707,000 in May to July 2026 and had been broadly flat since the start of the year. The Office for National Statistics also records that some smaller firms cited higher labour and operating costs as reasons not to recruit. Payroll employment had generally declined over the preceding two years, although the latest monthly estimate was broadly unchanged; the ONS rightly advises using several indicators because survey and administrative measures have different coverage and timing. (Office for National Statistics, 2026) ↗

That is not equivalent to a uniformly employer-friendly market. The Bank of England’s business intelligence points to easier recruitment for routine roles but persistent shortages in specialist skills and specific places. Its contacts reported using temporary recruitment and natural attrition to retain flexibility, while continuing to recruit in selected professional and IT/digital activities. (Bank of England, 2026) ↗ The practical consequence is important: an organisation can impose a general vacancy constraint while still being unable to staff the roles that protect service quality, deliver change or generate revenue.

**Sanctuary interpretation.** The useful unit of workforce planning in 2026–27 is not the total headcount budget but the capability portfolio. Leaders need to distinguish work that is genuinely scarce or commercially pivotal from work that can be simplified, augmented or retired. This matters particularly for SMEs, where a single missing technical lead, experienced supervisor or customer-facing manager can constrain growth more severely than a modest overall reduction in recruitment costs.

Why aggregate cooling can coexist with acute capability bottlenecks

Three connected indicators explain the apparent contradiction between softer hiring and continued workforce risk.

**First, vacancy levels describe volume, not substitutability.** The 707,000 vacancy figure is far below the post-pandemic peak and, outside the pandemic period, was last at or below this level in late 2014. Yet a lower total says little about whether an employer can replace a qualified electrician, an experienced production technician, a regulated-service professional or a manager able to lead a complex operational change. (Office for National Statistics, 2026) ↗ For employers, the more useful internal measures are vacancy-to-starter conversion, time to independent productivity, regretted attrition and reliance on a small number of critical individuals.

**Second, shortage pressure remains concentrated by sector.** In the 2024 Employer Skills Survey, 27% of vacancies were skill-shortage vacancies, equivalent to 250,500 vacancies; this was down from 36% in 2022, but plainly not eliminated. Construction had the highest incidence, at 45%, while education, manufacturing, primary industries and utilities also recorded rates around one third or above. (Department for Education, 2025) ↗ This is a warning against treating a softer national market as permission to defer all development spending. In infrastructure-linked, technical and regulated activity, recruitment weakness and skills scarcity can occur simultaneously.

**Third, pay pressure should ease unevenly rather than disappear.** Regular pay growth was 3.5% in April to June 2026, with private-sector regular pay growth at 2.8%; public-sector timing effects complicate any simple reading of the aggregate figure. (Office for National Statistics, 2026) ↗ The Bank’s September intelligence put the cumulative average 2026 pay settlement at 3.6%, with contacts expecting 2027 settlements to be similar or slightly lower. It also identified inflation, National Living Wage decisions, union activity and retention pressures as material uncertainties. (Bank of England, 2026) ↗

The trade-off is clear. A blanket pay or recruitment reset may help short-term cost control, but it can accelerate losses in the roles where external replacement is slow and expensive. Employers should target retention and development where capability is genuinely constrained, while avoiding automatic premium pricing for roles where labour supply has eased.

AI changes the entry route as well as the job

The most useful question for workforce planning is not whether AI will remove a particular number of jobs. It is which tasks are changing, what learning those tasks once provided, and who will take responsibility for quality when work is partly automated.

Bank of England contacts reported productivity improvements from AI in software development, finance, administration, customer service, professional services and content creation. They also reported reduced demand for some entry-level work, including document preparation, invoice processing and basic analysis. Some professional-services organisations had reduced graduate recruitment and demand for junior administrative staff. (Bank of England, 2026) ↗

This evidence should be used with care. Agents’ intelligence identifies business experience and emerging patterns; it is not a causal estimate of economy-wide job displacement. Nor does it establish that fewer junior tasks will necessarily mean fewer careers. AI can increase output, release time for higher-value work and create demand for data, analytical and AI-related skills. The same Bank evidence stresses that realised gains differ by organisation and depend on implementation, staff capability, human oversight and change management. (Bank of England, 2026) ↗

The sharper risk is a damaged capability pipeline. Routine junior work has often supplied repeated exposure to customer context, errors, exceptions and professional standards. If that work is automated without a replacement learning design, firms may save time today while reducing the supply of future supervisors, managers and specialists. This is especially relevant for smaller businesses, which often lack the capacity to absorb a failed hire or recruit experienced talent at a premium.

**Sanctuary recommendation.** Employers introducing AI should maintain a task-and-learning register: the task being redesigned; the knowledge and judgement previously acquired through it; the human assurance step; and the supervised practice, rotation or coaching that will now build competence. A smaller entry-level intake can be viable, but only if it has a deliberate route from AI-assisted task completion to independent judgement. Simply removing junior roles is a cost decision, not a workforce strategy.

The missing complement to technology investment: management and training

Technology adoption is not evenly distributed, and neither is the capacity to convert tools into productivity. In the 2024 Employer Skills Survey, 14% of employer sites reported using AI, rising to 24% of sites with 100 or more employees and 43% in information and communications. Among AI-using sites, 86% expected to embed it into processes and operations to at least some extent. (Department for Education, 2025) ↗ This is not a real-time census of adoption in late 2026, but it establishes an important structural point: size, sector and organisational capability shape who is equipped to make use of AI.

The limitation is that technology expenditure can be easier to approve than the less visible work around it: process mapping, data discipline, manager training, quality controls and redesigned progression. That matters because employer-funded capability formation is already fragile. Employer training expenditure was £53.0 billion in 2024 prices, 18.5% lower in real terms than in 2011; expenditure per employee was £1,700, the lowest level in the survey series. Only 59% of employers provided training in 2024. (Department for Education, 2025) ↗

For high-street firms and other small employers, the relevant question is not whether to imitate the most advanced adopters. It is whether a specific workflow has enough volume, repeatability and governance to justify redesign. A well-run change may reduce rework, improve customer response times and make scarce staff more productive. A poorly governed rollout can shift mistakes to customers, overburden managers and remove the junior tasks through which staff learn the business.

**Sanctuary interpretation.** Management capability is the bridge between AI adoption and local economic value. Skills providers, business-support organisations and employers should therefore package practical workflow diagnosis, manager development and measured task redesign together, rather than treating software access as the intervention.

Scenario outlook and the signals that would change the judgement

**Base case: selective hiring and uneven productivity gains.** Demand remains soft to moderate; employers use natural attrition, stricter replacement rules and targeted automation to protect margins. Hiring persists in revenue-critical, technical, regulated and digital roles, while pay settlements ease only gradually. AI produces visible benefits in selected workflows, but gains remain limited where process ownership, data quality and managerial capacity are weak. This is the most plausible scenario given flat vacancies, cautious business conditions and the Bank’s settlement evidence. (Office for National Statistics, 2026) ↗ (Bank of England, 2026) ↗

**Upside case: investment broadens capability demand.** Improving confidence translates into investment in digital transformation, infrastructure-linked activity and export-facing services. The Bank reports modestly improved investment intentions, selected strength in business services and manufactured exports, and continued technology expenditure including AI. (Bank of England, 2026) ↗ Recruitment would broaden, but chiefly for people who can implement change: technical trades, data and cyber specialists, project leaders and capable line managers.

**Downside case: cost pressure deepens the replacement freeze.** Renewed inflation or input-cost pressure prompts more employers to defer hiring, use contingent labour and reduce internal development. The Bank identifies inflation, National Living Wage decisions and uncertainty as risks to 2027 settlements, alongside weakness in consumer spending, construction and property-related activity. (Bank of England, 2026) ↗ In this scenario, junior opportunities decline faster than specialist shortages, worsening the longer-term pipeline problem.

These are directional scenarios, not numerical forecasts. The base case should be reconsidered if broad-based vacancy growth is accompanied by sustained payroll employment growth and a clear recovery in employer-funded training or apprenticeships. Conversely, renewed falls in payroll employment, a material decline in vacancies, greater dependence on temporary labour and further reductions in development activity would strengthen the downside case. Evidence that AI deployment is expanding, rather than reducing, structured entry-level recruitment at scale would also challenge the central pipeline concern.

A practical agenda: make hiring, redesign and development one decision

The appropriate response is neither a blanket recruitment freeze nor an undisciplined race to automate. It is a disciplined capability review focused on the workflows that most affect quality, revenue, resilience and future skills supply.

**For employers:** review five to ten consequential workflows over the next 90 days. For each, identify the customer or operational outcome; the scarce human capability required; the tasks suitable for augmentation or standardisation; the quality-control point that must remain human-accountable; and the effect on junior learning. Approve recruitment where it removes a demonstrable bottleneck or enables a defined opportunity. Invest in development where repeated external hiring is the more expensive substitute.

**For SMEs and local business-support partners:** concentrate on adoption foundations before tool proliferation. A practical offer combines workflow diagnosis, basic data governance, manager confidence and simple outcome measures such as cycle time, error rates, customer resolution and time to productivity. This is more likely to generate local value than generic AI awareness activity.

**For skills and employability providers:** work with employers to redesign entry pathways around supervised exposure to real decisions, auditable AI-assisted practice and transparent progression criteria. Employers may need fewer people to complete routine tasks, but they will still need people who can understand exceptions, exercise judgement and lead others.

The test for any workforce decision is therefore straightforward: does it improve current performance without weakening the organisation’s capacity to develop the people it will need next? In a subdued labour market, the employers that answer that question well should be better placed to protect productivity, service quality and sustainable growth.