Pathways to Work Innovation Fund: employer capability, not novelty, is the real test

The Fund’s real question is whether work changes

The Pathways to Work Innovation Fund arrives at a consequential point for the UK labour market. Its first application round has a deadline of 26 October 2026, and the Department for Work and Pensions (DWP) plans to invest up to £60 million over two years in ideas that help disabled people and people with long-term health conditions move towards, enter, stay and progress in work. The stated ambition is not only to fund delivery, but to generate approaches that can be adopted, replicated or scaled. Up to £10 million is allocated to the first round, with a further £50 million expected in a second round in early 2027 (DWP Fund guidance, 2026) ↗.

The scale of the underlying challenge makes that ambition important. The Keep Britain Working review reports 2.8 million people economically inactive because of health conditions. It also puts the disability employment rate at 52.8%, 29.5 percentage points below the rate for non-disabled people (Keep Britain Working review, 2026) ↗. These figures describe exclusion, but they do not tell commissioners which interventions will create durable change.

That distinction should shape how the Fund is judged. A pilot can look innovative because it uses a new digital platform, referral route or partnership label. Yet its practical value may be slight if the jobs available remain inaccessible, insecure or poorly matched, or if an employer needs permanent specialist support to retain a worker. Conversely, a less conspicuous intervention may be highly valuable if it enables managers to make adjustments confidently, redesigns an inflexible role, or gives a local enterprise network a repeatable way to recruit and retain people.

Sanctuary’s assessment is that the decisive test is not whether a proposal is novel in isolation. It is whether it changes the conditions through which people can obtain, keep and progress in good work—and whether employers and local institutions can continue using that change after grant funding ends. This is a test of work design, management capability and local implementation, not simply participant engagement.

Experimentation should begin with mechanisms that have some evidence behind them

There is a sound case for local experimentation. People’s health conditions, caring responsibilities, transport access, labour-market opportunities and previous experience of services vary sharply by place. A model that works in one locality may not reach people in another. But local variation is not a reason to treat every intervention as equally plausible.

A 2025 systematic review of UK interventions to reduce economic inactivity among people with long-term health conditions and disabilities identified 16 unique studies across eight interventions. It found the most robust evidence for classic Individual Placement and Support (IPS), identified conflicting evidence on sanctions, and highlighted significant gaps, including for deprived areas and the North of England (Haighton et al., 2025) ↗. An earlier overview of systematic reviews similarly found that vocational-intervention evidence varies substantially by population and intervention, with much evidence outside established supported-employment approaches rated low or very low quality (Levack and Fadyl, 2021) ↗.

The implication is not that the Fund should finance IPS alone. It is that applicants should state the mechanism they are improving, adapting or extending. For example, a project might test whether employment support embedded in a housing provider, community organisation, primary-care setting or business network reaches people and employers that mainstream provision misses. It might test practical adjustment brokerage for small firms, benefits and income advice integrated with job matching, or peer-led support that makes sustained engagement more likely. In each case, the proposal should identify the barrier, explain why its mechanism could address it, and distinguish its genuinely new component from standard employability activity.

There is a legitimate counterargument. Requiring evidence-informed design can become a demand to reproduce a service model that does not fit local communities. The answer is disciplined adaptation, not weak fidelity disguised as innovation: retain the features that appear important, document what changes locally, and test whether the variation improves outcomes for a clearly defined group. DWP’s Connect to Work business case treats adherence to IPS principles and the Supported Employment Quality Framework as associated with better outcomes and subject to structured assessment (Connect to Work Programme Business Case, 2026) ↗. That is useful operational guidance, although an implementation business case should not be mistaken for definitive causal evidence in every setting.

Employer capability is the missing unit of innovation

Too many employment interventions locate the problem almost entirely with the jobseeker: confidence, skills, search behaviour, disclosure or readiness. These factors matter, and specialist support can be essential. But the obstacle often sits within the workplace itself: inaccessible recruitment processes, rigid rotas, absence-management triggers, poorly designed workloads, uncertainty about adjustments, or line managers who lack the confidence to have constructive conversations about health.

DWP’s Fund guidance explicitly includes approaches that help employers support disabled people and people with health conditions (DWP Fund guidance, 2026) ↗. The Keep Britain Working review also identifies fear around disclosure, inconsistent employer support and structural barriers to entering and remaining in work (Keep Britain Working review, 2026) ↗. These are organisational problems. A better referral journey alone cannot resolve them.

A stronger project would therefore treat the employer as a co-user of the intervention and a measurable beneficiary. It would recruit a defined employer cohort, diagnose the points at which job design or management practice blocks retention, offer practical change support, and measure what remains after specialist input reduces. Relevant measures could include revised vacancy design, improved adjustment processes, managers’ confidence and capability, flexible-work options, and the ability to resolve issues without continuous external mediation.

This matters particularly for SMEs, high-street businesses and local social enterprises. Such employers may not have specialist human-resources capacity, yet their hiring and retention decisions are often made quickly by owner-managers or front-line supervisors. The credible offer is not an abstract request to demonstrate social value. It is help with ordinary operational concerns: hard-to-fill vacancies, avoidable turnover, loss of experienced staff through ill health and managers’ uncertainty about what practical support looks like. If inclusion is made operationally usable, the social and business cases can reinforce one another.

There is a trade-off. Deep employer work takes time and can reduce short-term placement volumes. But a pilot that generates fewer starts while building repeatable employer capability may have greater long-term value than one that produces more starts through intensive, time-limited brokerage. Fund assessment should make that trade-off visible rather than rewarding headline reach alone.

Participation depends on financial design as well as accessible applications

The Fund’s bid process includes measures intended to widen access. Applications are open to public, private and voluntary organisations; partnerships are permitted; a partnership directory is available; and materials are provided in formats including Easy Read, British Sign Language, large print, audio and braille (DWP applicant bid pack, 2026) ↗. These features matter, particularly where delivery depends on trusted community organisations and lived-experience leadership.

However, access to an application process is not the same as access to delivery. The bid pack states that payments will generally be made in arrears against agreed milestones and evidence. Mobilisation payments of up to 30% can be requested, but are not guaranteed and will be approved only in exceptional circumstances (DWP applicant bid pack, 2026) ↗. This creates a plausible working-capital risk for smaller organisations that may have strong local reach but limited reserves to cover payroll and delivery costs before reimbursement.

This is an inference from the funding model, not evidence that the Fund will exclude smaller providers. A well-capitalised lead organisation can carry risk fairly, and mobilisation support may be available where justified. Nonetheless, applicants should address the issue explicitly. Consortium agreements should set out payment terms for partners, responsibility for delayed claims, treatment of overheads, data responsibilities, decision-making rights and routes for lived-experience organisations to influence delivery rather than simply act as referral partners.

For funders and lead bidders, the principle is straightforward: if trust, cultural competence and local knowledge are material inputs to the model, they must be resourced as delivery capability. Otherwise, the programme risks selecting organisations best able to write bids and finance cash flow, rather than those best placed to co-produce credible service change.

Success requires an evidence-and-adoption contract, not just a job-start target

The Fund asks applicants to demonstrate potential for adoption, replication or scaling (DWP applicant bid pack, 2026) ↗. That requirement should lead teams to agree an evidence-and-adoption contract during mobilisation: a short, practical account of what they will test, what evidence they need, and who could own the model if it works.

First, define the population and the exclusion mechanism. “People with health conditions” is too broad to generate useful learning. A project should specify who it is designed for, which barriers are most material in that locality, and how people with lived experience informed the diagnosis. This does not demand artificial homogeneity; it requires enough clarity to understand for whom the model works and under what conditions.

Second, measure quality and durability as well as entry. Job starts should be recorded, but so should retention at 13, 26 and 52 weeks where feasible; contracted hours; whether adjustments are agreed and sustained; participant-reported job fit and choice; progression; and reasons for exit. These measures will not solve attribution problems, but they make it harder to confuse a short-lived placement with durable inclusion.

Third, measure employer change and delivery intensity. Teams should record changes to recruitment, role design, management practice and adjustment processes, alongside caseloads, specialist time, referral conversion, employer-engagement effort and unit costs. The question is not only whether the intervention worked, but what it took to work. Without this information, a later commissioner cannot distinguish a transferable model from a pilot held together by exceptional goodwill.

Fourth, name an adoption owner before delivery ends. Dissemination is not adoption. A published evaluation has value, but it does not amount to system change unless a credible organisation has the authority, budget, workforce and operating route to continue the model. That owner may be a local authority, integrated-care partner, mayoral authority, employer network, housing provider, further-education institution or social-enterprise consortium. The right answer will vary locally; the absence of an answer should be treated as a delivery risk.

What local partnerships should do now

The strongest applications and delivery plans will reduce ambiguity rather than accumulate fashionable features. Local authorities, VCSE organisations, employers, health partners and social enterprises should ask four hard questions.

First, which employer practice must change for this group to thrive in work? Second, which elements of the model draw on existing evidence, and which are genuinely being tested? Third, who bears financial, safeguarding, data and delivery risk across the partnership? Fourth, if the model succeeds, who will pay for it and operate it when grant funding stops?

For community organisations, answering these questions is also an enterprise-development task. Trusted local delivery is strengthened by a clear proposition for employers, realistic unit costing, appropriate governance, robust data practice and partnership terms that recognise communities as co-designers and delivery leaders. For employer partners, it means being willing to examine work itself—not simply offering a small number of placements.

The Fund should ultimately be judged by more than the number of pilots it announces. Its public value will lie in whether it produces a smaller number of approaches that participants can rely on, employers can use, communities can shape and commissioners can fund with confidence. That is the more demanding definition of social innovation. It is also the one most likely to reduce exclusion rather than merely document it.