At a glance
Client: An anonymised healthcare provider in Ireland
Service: Patient portal usability research and service design
Focus: Registration, identity confirmation, appointments, results, consent language and support
Participants: Illustrative mix of patients, including older adults, people with lower digital confidence and people using assistive technology
Outcome: A clearer, more accessible portal concept and a prioritised plan for fixing barriers before wider release
The challenge
The healthcare provider was preparing a portal for appointments, selected results and personal information. It also needed to explain consent and privacy choices clearly.
The main journeys were already built. The question was whether they made sense to patients who might be worried, unfamiliar with medical terms or using a phone. In healthcare, unclear instructions can leave someone unsure whether an appointment is confirmed, whether a result requires action or whether information has been shared.
The provider asked usability.ie to find these risks before wider release. Our role was not to approve clinical content or certify compliance. It was to study patient understanding and help the team respond.
What we needed to learn
We agreed a short set of research questions rather than attempting to test every screen.
- Could patients register without help and understand why identity confirmation was required?
- Could they tell the difference between requesting, booking, rescheduling and cancelling an appointment?
- Did the results area explain what was available, what was still pending and what to do next?
- Could patients understand consent and privacy choices without reading legal or clinical language several times?
- When something went wrong, could they find the right support route?
- Did the service remain usable with screen magnification, a screen reader, keyboard navigation and larger text settings?
Completion alone was not enough. Patients also needed to understand the consequence of each action.
How we approached the research
We began with a focused accessibility-led review of the key journeys. This was not a full technical accessibility audit. It identified likely barriers before patient sessions.
We then planned task-based sessions with an illustrative mix of older adults, people with different levels of digital confidence, family carers and assistive technology users. The sample was designed around the portal's patient population and the access needs relevant to its main journeys.
Each session used realistic but fictional personal and clinical information. Participants were not asked to reveal diagnoses, medical histories or login details. We observed them completing tasks such as:
- creating an account from an invitation;
- confirming identity using a one-time code and date of birth;
- checking whether an appointment had been booked;
- changing an appointment time;
- locating a newly available result;
- deciding what a result status meant;
- reviewing a consent choice; and
- finding help after an unsuccessful identity check.
Participants explained what they expected before acting and what they believed had happened afterwards. A task can appear successful in analytics while the patient leaves with the wrong understanding.
We also reviewed increased text sizes and keyboard navigation. In this illustrative plan, a screen reader user tested form labels, reading order and status messages. Accessibility issues were recorded with the wider findings.
What patients helped us see
Registration asked for trust before it had earned it
The invitation used the provider's name, but the registration page moved quickly into identity questions. Some participants were unsure why they had been sent to a different web address or why the service needed their date of birth again.
The issue was not simply a missing sentence. The journey did not build a clear chain of trust. It needed to explain who operated the portal, why identity confirmation protected the patient and what information would be available after registration.
We recommended explaining this before the first personal-data field, using the provider's familiar name consistently and offering a way to verify the invitation independently.
Identity confirmation errors did not explain recovery
When the illustrative test account failed an identity check, the message said the details did not match. It did not say which details could safely be checked, how many attempts remained or whether a patient should contact the clinic or the portal support team.
The revised design separated safe self-checks from issues requiring staff support. It preserved information already entered where appropriate and placed the relevant phone number beside the error.
Appointment language reflected the system, not the patient's task
The portal distinguished between a requested appointment, a provisionally offered slot and a confirmed booking. These states were meaningful to the scheduling system, but their labels were too similar for patients.
We rewrote statuses around patients' questions: "Do I have an appointment?", "Do I need to do anything?" and "When will I hear back?" Dates, locations and next steps appeared together. Cancellation language made the consequence explicit.
Results needed context, not interpretation
The results area presented document titles and publication dates. Participants did not always recognise the clinical name of a test, and a "new" badge sometimes appeared more urgent than intended.
Usability.ie did not rewrite clinical advice. We helped present approved explanations more clearly. The proposed design distinguished a result being available from it being reviewed, and stated what would happen next. Interface language must not guess at the meaning of a clinical result.
Consent text mixed several decisions together
The consent screen attempted to cover portal terms, privacy information, notifications and information sharing in one long passage. Some participants selected the main checkbox simply to continue.
We separated required acknowledgements from optional choices. Each received a plain-language summary, further detail and an explanation of what would change if the patient said no.
Support appeared too late
The original help link sat in the footer. Patients who were blocked during registration did not consistently find it, particularly on a phone or at high zoom.
The revised concept placed help at the point of failure and distinguished technical, clinical and appointment queries. Each channel stated its opening hours and expected response time, subject to provider confirmation.
Turning findings into changes
The final research playback was organised by patient risk and service impact, not by screen. The team received:
- a prioritised findings report with evidence from observed sessions;
- annotated journey maps showing where confidence dropped;
- plain-language recommendations for registration, status and consent content;
- accessibility findings with reproduction steps;
- short clips for internal review, subject to participant consent;
- an agreed list of changes to test in the next prototype; and
- unresolved clinical, legal and operational questions assigned to the appropriate owner.
We recommended retesting identity recovery, appointment status and results messaging with patients after revision.
Illustrative results: sample reporting only
- Sample result: Successful unaided registration increased from 58% to 87% in a second round of testing.
- Sample result: The proportion of participants who correctly understood whether an appointment was confirmed increased from 61% to 94%.
- Sample result: Identity-confirmation support contacts fell by 24% during the first six weeks after release.
- Sample result: Median time to reschedule an appointment fell from 3 minutes 40 seconds to 1 minute 55 seconds.
- Sample result: The number of patients who could identify the correct support route after a failed identity check increased from 5 of 10 to 9 of 10.
- Sample result: Critical and serious accessibility issues in the tested journeys fell from 11 to 3 after remediation and retesting.
- Sample result: Patient confidence, measured on a five-point post-task scale, increased from an average of 3.1 to 4.3.
Published figures should state sample size, measurement period and concurrent changes. A percentage without its denominator or context can overstate modest evidence.
What this work demonstrates
Patient portal research tests whether patients know where they are, what information is requested, what an action means and what happens next. Usability.ie can support early journey reviews, patient research planning, moderated testing, accessible sessions, plain-language recommendations and retesting. Clinical decisions, legal advice and formal accessibility certification remain with qualified specialists.
Have a similar usability question?
Tell us which journey matters, what evidence you already have and what decision the research needs to support. We will recommend a proportionate audit or research approach.