<p>The meeting room at the municipal employment centre is full before the demonstration begins. On the screen is a prototype that promises to organise appointments, workshops, and follow-up for people looking for work. The supplier opens a dashboard of blue bars: minutes connected, forms submitted, responses to notices, and tasks completed. A caseworker asks whether the system will distinguish someone who spends an afternoon preparing an application from someone who presses refresh five times. The first answer is that both actions leave a trace. That is where the problem begins: recording activity is not the same as understanding the work a person is doing.</p>
<p>The tool is public because it shapes a public-service relationship, even if a company wrote part of the code and people access it from private phones. It may affect the support someone receives, which alert gets priority, or when a professional decides to call. Designing it without informed participation turns users and frontline staff into data sources rather than interlocutors. Doing it well does not mean collecting opinions at the end of a contract. It means enabling those who will carry the consequences to understand what will be measured, challenge its purpose, and alter decisions before they become routine.</p>
Making the context clear#
<h2>The difference between being visible and doing work</h2>
<p>A presence counter is appealing because it looks neutral. It is easy to aggregate, fits on a dashboard, and makes weeks comparable. Yet in an employment service it can punish exactly the tasks that require care: reading a vacancy slowly, asking for help to interpret a requirement, caring for a relative before an interview, or deciding that a course is a poor fit. It also ignores that not everyone has reliable connectivity, a personal computer, or a quiet time slot. When a screen turns those limits into low participation, it converts material inequality into an administrative label.</p>
What changes in practice#
<p>The first design conversation should separate three questions that are often fused together. What does the person need in order to move forward? What does the team need to know to provide good service? What data is convenient to collect? The third question cannot answer the first two by default. If the aim is a useful appointment, it may matter more that someone understands the available options and chooses a realistic next step than how many clicks they make. If the aim is to spot a blockage, a free-text message saying “I do not know how to document my experience” may reveal more than three days without a login.</p>
<p>Informed participation supplies knowledge that no requirements document contains. A newcomer can explain why an automated notice written in legal language is not understood. A careers adviser can show that low activity sometimes means a person has found temporary work, not that they have abandoned the process. A disabled worker can point out that a mandatory video call creates a barrier the team did not see. These contributions are not anecdotes to file away as evidence of listening; they are evidence that should change the rules of the tool.</p>
Deciding with information#
<h2>Inform people before asking for consent</h2>
<p>An honest invitation does not begin with an acceptance box. Before showing the prototype, the council should provide a short plain-language sheet explaining which decisions are still open. For example: which events the platform will record, how long they will be retained, who will see activity profiles, whether a data point can affect appointment priority, and what option exists for using the service offline. It must also name what is not yet known. Saying “we are still deciding whether the dashboard will classify participation levels” opens a discussion; hiding that choice behind a promise of efficiency closes it.</p>
Reviewing and adjusting#
<p>The information must make it possible to reject a proposal without losing access to the service. If taking part in a design workshop requires accepting detailed observation of personal use, the process selects people with fewer objections or less room to voice them. In-person sessions, phone calls, accessible forms, and paid conversations are all appropriate when asking for expert time from people in a financially fragile position. A variety of channels is not an inclusive decoration: it changes which risks are found and who has authority to name them.</p>
<p>Consequences matter as much as data fields. Rather than asking “do you authorise activity logging?”, the team can show three fictional screens: one in which a professional sees a weekly summary, one in which they receive an absence alert, and one in which no metric enters a service decision. Participants can then identify which interpretation they fear, what explanation they would need, and which boundary they would regard as fair. That conversation prevents consent from shrinking into an abstract word detached from its actual effect.</p>
<h2>A pilot that can change the product</h2>
<p>The best test is not a satisfaction survey after launch. It is a small pilot with an end date and explicit conditions for stopping. The group might include ten users with different paths, two local associations, advisers, data-protection staff, and someone outside the project whose role is to ask uncomfortable questions. For six weeks, every disputed decision should enter a log: what the team proposed, what objection arose, what was changed, or why it was rejected. The log should not reveal personal stories; it should make the institution's reasoning visible.</p>
<p>Suppose the pilot shows that reminders sent at eight in the morning create more missed appointments than attendance. The relevant change would not simply be to move a notification. The team should ask whether the system was measuring an imagined availability, whether the messages pressured people with irregular shifts, and whether the response metric was being treated as a signal of commitment. A responsible correction records that chain. Then, if someone proposes restoring the automatic alert six months later, there will be evidence of why it was limited.</p>
<p>Frontline staff need the same right to disagree as users. If an adviser sees that the dashboard pushes staff to close cases because activity appears low, they should be able to mark the alert as misleading without asking the supplier's permission. That power requires training, a review route, and protection against poorly designed internal targets. Final responsibility cannot dissolve into the sentence “the tool recommends it.” An algorithm may suggest; a public institution must be able to explain, correct, and, when needed, decline the suggestion.</p>
<h2>Measuring a service without measuring obedience</h2>
<p>There are more useful indicators than screen time. The centre can examine whether people understand their options, secure an appointment within a reasonable period, correct a data point, receive an answer when they raise an exception, and encounter exclusion through the digital channel. These indicators are not perfect, but they shift attention from visible obedience to the quality of a public relationship. They also require admitting that a good outcome may be someone using the platform less because they solved their need through another route.</p>
<p>For such measurement to be credible, every metric needs a public card: its purpose, source, potential bias, human owner, retention period, and decision it must never make on its own. The card should include an example of misuse. “The number of logins will not be used to lower service priority” is far clearer than a general statement about responsible use. A person without technical knowledge should be able to understand it and request review without writing a legal brief.</p>
<p>The next practical action is to convene a design session before procuring the final version. The team should bring options rather than a closed solution, explain what can change, document disagreement, and publish a response to each significant proposal. Informed participation does not guarantee unanimity, but it stops efficiency from becoming an excuse for decisions behind closed doors. A public tool earns legitimacy when affected people can influence how work is defined, what counts as a signal, and where a human exit must always remain.</p>
