When Services Become Factories
How public digital services came to operate like production lines
If you’ve been following this series so far, you’ll remember that last month I described how systems such as Probation Services and GOV.UK Verify exposed the limits of transactional service models. These failures are often explained as implementation problems - programmes that were poorly delivered or systems that were badly designed.
But the pattern they reveal is deeper than that.
The systems themselves are organised around a logic that prioritises procedural flow over human complexity.
Once you begin to look through that lens, the architecture of many public digital services starts to feel strangely familiar.
They behave like production lines.
The logic of the production system
Modern manufacturing is built on a deceptively simple idea: work should move through a system in a smooth and predictable flow.
The model most often associated with this approach is the Toyota Production System, developed in Japan in the second half of the twentieth century. Toyota’s method transformed industrial manufacturing by focusing on the elimination of waste and the maintenance of continuous production flow.
One of its core principles is Just-in-Time production. Each stage of the process produces exactly what the next stage requires, exactly when it is needed. Parts arrive at the moment they are required. Workers follow standardised sequences of tasks. Products move through a defined set of stages until the finished item leaves the line.1
The objective is not flexibility. It is flow.
When variation appears - a faulty component, a missing part, a delay in supply - the line stops until the problem is resolved. The system depends on predictable inputs moving through predictable processes.
In manufacturing this logic works remarkably well. Cars and components behave within known tolerances. Tasks can be standardised. Production can scale.
Now compare that structure with how many public digital services operate.
Applications must follow a fixed sequence of forms. Evidence must be provided at specific points in the process. Eligibility rules determine whether a case progresses to the next stage. Automated systems perform tasks that once required human judgement.
The language is different, but the structure is strikingly similar.
Standardised work becomes standardised applications.
Quality control gates become eligibility checks.
Production flow becomes case processing.
A person entering the system begins to resemble a unit of work entering a production process. How is this human-centred-design?
Universal Credit - a processing pipeline
in the United kingdom, the Universal Credit system offers a clear example of how this logic appears in practice.
Introduced in 2013, Universal Credit merged six existing welfare benefits into a single system intended to simplify administration and encourage employment. Claims are managed through an online account where applicants provide information about their circumstances, verify their identity and report changes in income or employment.2
Behind that interface sits a carefully structured administrative process.
First, a claimant completes an online application detailing their personal information, housing situation and employment status. Identity verification is then carried out through a digital identity system or, where that fails, through in-person verification. Supporting evidence such as tenancy agreements or proof of earnings must be submitted before the claim progresses.
Once these steps are complete the claim enters what the Department for Work and Pensions calls the assessment period. During this period the system calculates entitlement based on reported income and household circumstances. At the end of the assessment period the first payment is scheduled.
For most claimants this results in a delay of around five weeks between the initial claim and the first payment.3
From an administrative perspective the structure is logical. Identity must be verified, entitlement calculated and evidence confirmed before public funds are distributed.
But the design also reveals the deeper architecture of the service.
Each stage produces the information required for the next stage.
Identity verification enables eligibility checks.
Eligibility checks enable payment calculations.
Payment calculations trigger payment.
The claimant moves through a fixed sequence of steps that cannot easily be rearranged.
The system functions as a linear processing pipeline.
This structure has been the subject of sustained scrutiny. The National Audit Office concluded that the rollout of Universal Credit had been “slower and more complex than expected”, and that the Department for Work and Pensions had “not yet demonstrated that the programme would deliver value for money.”4
Parliamentary committees have also highlighted the impact of the five-week waiting period. The House of Commons Public Accounts Committee reported that the delay had contributed to financial hardship for some claimants and raised concerns among local authorities and housing providers about rising rent arrears during the rollout of the programme.5
Researchers analysing administrative data have similarly identified correlations between the introduction of Universal Credit and increases in landlord repossession actions in some areas during early phases of the policy’s implementation.6
The debates surrounding the programme are complex. But the structure of the system itself reveals something important.
Universal Credit is not simply a welfare payment.
It is a large-scale administrative processing system.
The friction between systems and lives
The logic of the system becomes most visible when it encounters the unpredictability of everyday life.
Factories depend on controlled variation. Each component entering a production line behaves broadly the same way. If a part arrives damaged or missing, the problem can be isolated and resolved.
Public services do not operate under those conditions.
People arrive with circumstances that are unstable, incomplete or difficult to categorise. Someone applying for support may be navigating unemployment, illness, housing insecurity or caring responsibilities at the same time. Their situation may change several times before an administrative process reaches its conclusion.
Yet the system still requires them to move through a predefined sequence of steps.
If evidence is missing, the claim cannot progress.
If information changes, the process may restart.
If identity verification fails online, additional checks must be completed before the system continues.
From the perspective of the administrative system this behaviour is rational. It protects the integrity of the process and ensures that decisions remain consistent.
From the perspective of the person navigating the service it can feel arbitrary and inflexible.
The friction does not arise from poor intentions.
It arises from a structural mismatch between human complexity and procedural flow.
Why services begin to resemble factories
These patterns are not accidental.
They emerge from the institutional conditions under which public services are designed and funded.
Public spending in the United Kingdom is allocated through multi-year spending reviews that set departmental budgets against policy commitments. Departments must demonstrate that programmes can deliver those commitments with financial control and administrative consistency.⁷
Digital systems designed around standardised processes make those outcomes easier to measure and easier to audit.
Online forms capture information in consistent formats. Eligibility rules can be encoded directly into software. Automated workflows allow millions of transactions to be processed with limited human intervention.
The system becomes easier to scale.
Scale, accountability and fiscal control quietly push services toward structures that resemble production systems. Standardisation makes decisions predictable. Procedural workflows make outcomes measurable. Automated checks make compliance auditable.
None of these goals are unreasonable.
In fact they mirror the principles that made modern manufacturing so effective.
But the consequences are significant.
When services are designed primarily to maintain smooth flow through administrative processes, the system begins to optimise for its own operational stability rather than the complexity of the lives moving through it.
The structural tension
The factory floor is one of the most successful organisational inventions of the Industrial Revolution.. It allows complex work to be performed at extraordinary scale with remarkable efficiency.
But this is 2026 and factories are designed to produce identical objects.
Public services exist to support people whose lives rarely follow identical patterns.
When the systems built to support them begin to resemble production lines, the tension is not surprising. It is structural.
Designers of services need to acknowledge that this is not a question of whether the factory model exists in public services. It clearly does.
From here on out I’ll be posing a far more interesting question - what kinds of services become possible when systems are organised around human relationships rather than production flow?
Toyota Motor Corporation – Toyota Production System.
https://www.toyota-global.com/company/vision_philosophy/toyota_production_system/
Department for Work and Pensions – Universal Credit: How it Works.
https://www.gov.uk/universal-credit
House of Commons Library – Universal Credit: The Wait for First Payment.
https://commonslibrary.parliament.uk/research-briefings/cbp-8299/
National Audit Office – Rolling out Universal Credit.
https://www.nao.org.uk/reports/rolling-out-universal-credit/
House of Commons Public Accounts Committee – Universal Credit: Progress Review.
https://publications.parliament.uk/pa/cm201719/cmselect/cmpubacc/576/576.pdf
Analysis of housing security impacts during Universal Credit rollout.
https://www.sciencedirect.com/science/article/pii/S0016718522002111
HM Treasury – Spending Review Framework.
https://www.gov.uk/government/publications/spending-review-2021-documents