The appointment looked like a simple improvement: choose a clinician on a phone and receive an SMS code to confirm the time. In the city, it reduced queues and calls. In a village with unreliable coverage, it became an endurance test. Marta needed to renew a prescription. The first message never arrived. The second appeared after the window for entering it had expired. The call shop where she sometimes gets signal closed at noon; the health centre's line was busy, and the person at reception could no longer book directly because the screen required the code.
No predictive model explicitly decided to exclude Marta. Yet the innovation distributed the chance of reaching a consultation unevenly. That is what gets lost when infrastructure means only a mast, an app, or a network. Infrastructure also includes time to retry, the cost of travelling, privacy on a shared device, the presence of someone who can solve an incident, and the right to use a route that does not depend on stable connectivity. When a public service becomes quicker for people with continuous coverage and more uncertain for people without it, this is not a minor inconvenience. It is an access condition that creates inequality.
When an average conceals a place#
Teams commonly validate a service with aggregate measures: completed appointments, cost per transaction, average waiting time, satisfaction among people who answer a digital survey. Those figures can genuinely improve while also hiding a quiet withdrawal of access. A person who does not receive the SMS may not appear as excluded; they appear as an abandonment. Someone without credit, battery, or privacy to finish the process may be counted as inactive. The statistic then describes behaviour imposed by infrastructure as though it were an individual preference.
Teams commonly validate a service with aggregate measures: completed appointments, cost per transaction, average waiting time, satisfaction among people who answer a digital survey.
Measuring by territory changes the question. It is not enough to know that 92 percent of codes are delivered. An organisation needs to see where delivery fails, how long each resend takes, how many people switch channels, how far they travel to fix it, and how many appointments never happen after a third attempt. These data should be read alongside coverage, transport schedules, language, disability, caring responsibilities, and shared-device use, without turning living conditions into stigmatising labels. The purpose is not to classify villages as troublesome. It is to identify which design makes residents bear an extra cost.
Good territorial measurement places short accounts beside the numbers. A patient who walks to a hill to receive a message supplies information that a delivery rate does not contain. A receptionist who says every morning that the screen offers no alternative code identifies an organisational bottleneck. Such observations need a dated review, an owner, and authority to make a correction; they should not survive merely as anecdotes from a workshop.
Equivalence is not a help link#
Many organisations believe the issue is solved because a website has frequently asked questions or a generic phone number exists. That does not necessarily make an equivalent alternative. An alternative route works only when it can complete the same transaction, within the same deadline, without demanding extra data, imposing disproportionate cost, or delivering worse treatment. If a call only tells someone to return to the app, the alternative is decorative. If in-person staff can take a note but cannot confirm the appointment, it is not equivalent to the digital channel either.
For a health appointment, genuine equivalence may combine several options. The centre should be able to book at reception without first making the person prove that the SMS failed. An answered telephone should confirm date and time, rather than merely offer guidance. There may be a voice code, a short letter, or permission for a relative to assist without disclosing the clinical reason. The right mix differs by place, but every option shares one test: the administrative outcome must be the same, and no one should have to prove they deserve an exception.
Privacy is part of equivalence as well. Asking a neighbour to receive the SMS may solve an immediate problem and create another: it reveals a health visit to someone who should not know about it. Requiring use of an employer's phone or a call shop is not neutral. Design must provide channels; it cannot outsource the duty to repair its own fragility to informal networks.
Test the service where it fails#
The most useful test is not conducted in an office with steady wifi. It happens with intermittent signal, an older phone, limited battery, and a shift that does not permit six attempts. Before deploying a change, a team can simulate a late message, a missed call, a shared device, and a bus route that runs only twice a day. This is not about performing hardship or turning rural residents into a laboratory. It is about recognising that a public service cannot call a condition universal when it has tested it only in privileged places.
That testing needs people who work locally. Health staff, community workers, libraries, associations, and municipal offices can identify schedules, languages, and journeys that a technical dashboard cannot see. Their participation should be paid and able to alter requirements, not a consultation that decorates a project presentation. When they explain that the call shop closes before a cleaning shift ends, the system needs to answer with a specific route, not a promise to study the issue.
It is also useful to set an intervention threshold. If code delivery or appointment completion falls persistently in an area, a human review starts and direct booking becomes temporarily available. This is not a territorial privilege; it is maintenance of the promised service level. Just as an organisation repairs a ramp that no longer permits entry to a building, it repairs a digital flow that no longer permits access to a consultation.
Governing a previously unseen cost#
The final issue is not whether digital tools should exist. When designed well, they can save journeys and increase autonomy. The issue is who pays when the conditions those tools assume fail. If the cost always lands on people with weaker signal, less time, or less financial margin, efficiency has been funded by invisible inequality.
A responsible institution can publish a straightforward view: appointments started and completed by area, retries, use of each channel, repair delays, and cases where an alternative route achieved the same result. It should pair this with a periodic case review and changes explained in plain language. There is no need for maps that identify households or grand promises. A person needs to be able to see that their difficulty counts and that an entry route works today.
Marta does not need a lesson in digital transformation. She needs her consultation to be accessible even when the message does not arrive. The next step for any appointment team is to visit an area with unreliable coverage, complete the process under real conditions, and document every point where effort shifts to the patient. Then it should enable a phone or in-person route with identical booking power and measure each month whether abandonment clusters in that territory. Innovation becomes fairer when it stops asking only how much it accelerates and starts answering for whom access remains possible.
Test the route when the signal fails#
An equivalence promise needs a failure exercise, not a logo in a service manual. Once a month the clinic can ask a staff member to book the same appointment with a weak signal, an expired handset and no private email address. The test should record when the person reaches a human being, whether the available slot matches the online offer, and how much information they must repeat. A telephone line that merely says “use the website” is not an alternative; it is a second way to encounter the same barrier.
The results belong in service planning, alongside waiting times and completed bookings. If people in one area abandon the SMS stage more often, the remedy may be a staffed callback window, a local partner able to confirm appointments, or a paper notice with a number that works outside office hours. Naming the territorial pattern prevents low connectivity from being misread as low demand. It also gives managers a reason to protect the non-digital route when efficiency budgets become tight.
