What I Wish I Had Known Earlier About Designing for Public Services

From the outside, designing public services looks like ordinary service design with more paperwork attached. It is not. The differences are structural, and most of them only become visible once you are inside a project and something that should obviously have worked does not. These are the things I wish someone had said plainly at the start.

The Person Using the Service Cannot Walk Away

In commercial service design, a bad experience has a natural corrective. People leave, and the organisation eventually notices the leaving. Public services rarely have that mechanism. Someone applying for financial support cannot take the application to a competitor, and someone waiting on a diagnosis cannot choose a different health system because the phone line was engaged. Exit is mostly unavailable, so dissatisfaction accumulates quietly rather than appearing in a number anyone is watching.

This changes what the measurements mean. Completion rates stay high in services people cannot abandon, and high completion can conceal an enormous amount of effort: four phone calls, a day off work, a form filled in by a relative. I have learned to distrust any public-service dashboard that only reports whether things finished. The question worth asking is not whether people completed the process, but what it cost them to complete it.

It also raises the weight of every design decision in a way that took me too long to feel properly. If a retail checkout is confusing, the cost is a lost sale. If an eligibility form is confusing, the cost may be someone going without money they were entitled to. The interface problem is the same. The consequence is not.

Rules Are Part of the Material, Not an Obstacle to It

Early on I treated statutory constraints the way I had treated technical constraints in commercial work: as obstacles to be negotiated down. That is usually the wrong posture. A statutory requirement often exists because someone was previously harmed, or because a discretionary decision was previously made badly, and the rule is the residue of that. Understanding why a rule exists tends to be more productive than campaigning to remove it.

It also changes what good design looks like. Proving eligibility is the clearest case. The requirement to verify entitlement may be genuinely non-negotiable, but almost everything about how verification happens is designable: what evidence counts, how many separate times a person must supply it, whether one part of the organisation can see what another already holds, and what happens to someone who cannot produce a document precisely because of the circumstances that made them eligible in the first place.

The skill is separating the rule from its current implementation. Most of the pain usually sits in the implementation.

Booking a medical appointment shows the same pattern from a different angle. Clinical safety requirements, staffing rules and referral protocols genuinely constrain what can be offered and to whom. Almost none of them require the booking line to open at the hour most people are at work, or the confirmation to arrive without a clear statement of what happens if the person cannot attend. When a service feels immovable, it is worth checking how much of that immovability is statute and how much is habit that has been mistaken for statute.

Policy Intent and Lived Experience Are Different Objects

A policy is written as a description of an intended outcome. A service is what that description turns into once it has passed through budgets, legacy systems, staff training, letter templates, call scripts and the discretion of whoever picks up the phone. The distance between the two is not normally caused by bad faith. It is caused by the fact that intent is written in one vocabulary and delivery happens in another, and very few people are fluent in both.

The gap shows itself in small, unglamorous places. A policy intends a decision within a stated period, and the letter communicating that decision is written in language requiring a further phone call to interpret. A policy intends support to be accessible, and the accessible route requires an appointment only available during working hours. Nobody chose these outcomes deliberately. They emerged from the joins between one intention and another.

What I wish I had understood earlier is that documenting the gap carefully is itself a significant part of the work. Policy teams are frequently unaware of how their intent lands once it reaches the front line, and a precise, unhostile account of what happens between the rule and the person can move things that argument on its own cannot.

One Service, Several Organisations, No Single Owner

The hardest structural feature of public services is that what a person experiences as one service is often delivered by several organisations, none of which owns the whole of it. Someone seeking help may pass through a national agency, a local authority, a health provider and a voluntary organisation, each with separate funding, governance, data and definition of success. There is no single owner to convene, because in a meaningful sense there is no single service. There is a set of organisations whose outputs a person has to assemble themselves.

There is genuine research on what working across that boundary requires. Loban and colleagues, in a 2021 cross-case analysis published in Health Science Reports, examined multi-stakeholder partnerships in primary health care in two Canadian provinces, involving clinicians, health authorities, community organisations and patients. What matters here is not a figure to quote. It is that partnership across organisational boundaries is treated as a subject with its own conditions and its own difficulties, rather than as something that follows automatically once everyone agrees the goal is worthwhile.

Practically, this means a large share of the work is not design at all. It is establishing who has authority to change each part, discovering that nobody has authority over all of it, and building enough shared understanding that several organisations move in a compatible direction without any single instruction telling them to.

The People Most Affected Are the Hardest to Reach

There is an uncomfortable inversion in public-service research. The people who most need a service to work well are frequently the people least likely to appear in a research sample. They may have no settled address, no reliable internet access, very little spare time, low trust in the institution asking the questions, or entirely reasonable grounds for avoiding contact with it. Recruiting through the usual channels reaches the people who were already reachable.

That inversion quietly biases everything downstream. If a sample skews towards people who got through the process, the redesign optimises for people who were already going to succeed. The failures stay invisible, because failure in a public service often looks like absence rather than complaint: an application never started, an appointment never booked, a letter never answered, a person who concluded some time ago that the system was not for them.

The response I have found most useful is to go where people already are and to work through organisations they already trust: a community group, a support charity, a frontline worker who sees the same problem forty times a week and can describe it better than any survey. It is slower, and it is less tidy, and it is worth it. Absence is data, and it is usually the data most likely to be missing from the room.

What I would tell myself

None of this makes public service design harder to justify. It makes it more consequential. The absence of exit, the weight of statutory rules, the distance between policy and delivery, and the diffusion of ownership all point at the same underlying fact: outcomes in public services are produced by an arrangement of actors and rules, not by any single organisation's intentions, however good those intentions are.

If I could send one instruction back to myself, it would be to spend the first weeks of any public-service project working out who is permitted to change what, and who never sees the consequences of their own decisions. Almost everything else follows from knowing that.