Patient Portal Adoption: Help Patients Complete One Useful Task

Decide what adoption should accomplish

"Increase portal use" does not explain what a patient should be able to do. Choose an administrative task that the existing portal supports, such as submitting an approved contact-details form and receiving confirmation that it arrived.

Write the promise in ordinary language: "You can send this form through your account and see confirmation that the office received it." State separately if staff must review the information before it changes the main record. A submitted request and an accepted update are different outcomes.

Confirm that the task is useful to the patient, that the necessary information is available and that the receiving team can handle it. There is little value in promoting a form that creates an unattended inbox or a second request for the same information.

The broader digital front door strategy covers the relationship between portals, phones and other access channels. This guide concentrates on one journey inside an existing portal, rather than selecting a new platform.

Use national access data for context

ONC's analysis of the 2024 Health Information National Trends Survey reports that 65% of individuals nationally were offered and accessed their online medical record or patient portal at least once in the previous year. That survey measure is different from completing a particular clinic task. See the ONC data brief on patient portal access and use, 2024.

Do not turn that national result into a local task-completion target. Your denominator might include newly invited patients at one practice over two weeks. The survey concerns a broader population, multiple possible portals and a much longer access window.

For local improvement, establish the current result using your own defined task. Record who was offered the portal, what they were asked to do and how long each person had to complete it. Keep the difference between a missing event and a known unsuccessful attempt visible.

Build a funnel that explains the next investigation

Use one cohort and one task version. A patient with several visits should not appear as several newly invited people unless the measurement explicitly concerns visit-level opportunities. Existing account holders and first-time enrollees may need separate views because their starting conditions differ.

The following is an original operational worksheet. Use only events your authorized systems can establish; mark a step unobserved when the data does not support it.

A delivery receipt does not prove the patient read an invitation. A page opening does not prove the patient understood the instructions. Use those events to locate a question, then investigate it with appropriate observation or feedback.

Do not force incomplete telemetry into a perfectly ordered funnel. Someone may activate through an in-office route even when the original message has no delivery event. Preserve that alternate path rather than inventing a missing event to make the chart fit.

For an appointment-booking project, use the separate patient self-scheduling guide. Booking rules and transaction confirmation require their own design beyond portal access.

Work through a complete cohort

Hypothetical example: a practice invites 200 patients without active accounts to submit one administrative form. It measures each patient 14 days after their invitation. The period is an illustrative observation window, not a recommended deadline for patients.

In this example, 160 invitations have confirmed delivery. Of those patients, 120 activate access, 90 start the form and 72 submit it with a receipt. For simplicity, all downstream events follow that observed invitation path; the example has no alternative activation route.

The mutually exclusive outcomes at the checkpoint are:

Activation is 120 / 200, or 60%. Portal task submission is 72 / 200, or 36%. Among patients who activated, submission is 72 / 120, also 60%. Among those who started the form, submission is 72 / 90, or 80%.

The two 60% figures answer different questions. Labelling both "engagement" would hide the difference between establishing access and completing work after access.

Suppose 20 of the 72 submitters received approved staff assistance and 52 completed independently. Both count as portal submissions. Report the assistance separately so a successful supported journey does not disappear, and so staff effort is visible.

The next investigation is not automatically "send more reminders." Eighty patients have no activation event, including 40 without confirmed invitation delivery. Find out what is known about those routes and access attempts before attributing the gap to patient motivation.

Watch the journey before rewriting the invitation

Rehearse with a synthetic account, then use an authorized, voluntary approach to observe patient usability where appropriate. Start with the actual invitation and follow its link on the relevant device. Testing only an already-open desktop account skips the access steps the project is trying to improve.

Ask the participant to describe what they expect to happen next. Observe where the task becomes unclear, where a message changes language, or where they leave to seek help. Record the observed obstacle separately from your explanation of it.

An original test card can use five prompts:

1. Find the invitation and explain which office and task it concerns.
2. Reach the intended account through the approved access process.
3. Find the administrative action without a staff member pointing to it.
4. Complete or deliberately save the task, as supported by the actual portal.
5. Show the confirmation and explain what still needs to happen.

Test an invalid field, an interrupted session and a return to an unfinished form when those behaviors are relevant. The result should identify a specific repair: unclear field instruction, inaccessible confirmation, unsupported save behavior or another observed problem.

For example, a form might submit correctly but return the patient to a home screen with no visible acknowledgment. Rewriting the invitation would not address that uncertainty. First establish whether the existing system supports a clearer receipt or a different approved completion message.

For language support, inspect the whole task, including validation and confirmation. The multilingual outreach guide provides related channel context. Do not assume a translated invitation means the destination journey supports the same language.

Make assistance part of the operating plan

ONC's Patient Engagement Playbook discusses in-office enrollment support and clear instructions. It also describes proxy access for caregivers, with its own permission levels, rather than use of the patient's credentials. Apply your organization's approved access arrangements when offering help.

Give assistance a named team, an available route and a clear stopping point. Staff should know whether they can explain navigation, help with the approved account-recovery process or route an access question to another owner. They should not improvise a verification bypass to improve the completion rate.

Use the portal's authorized caregiver or representative access process when it applies. Do not ask patients to share passwords with staff or caregivers as an onboarding shortcut. Questions about permitted access should go to the organization's responsible team.

Offer an appropriate alternative when the portal route does not work for the patient. Record whether the underlying administrative task was completed through that route. That is a successful service outcome even though it is not independent portal completion.

Measure the support work you are asking staff to perform. A campaign inviting more patients needs enough receiving capacity to answer access questions and handle submitted forms. A help number that sends patients into an unowned callback list simply moves the problem.

Repair one failed step and retest it

Choose a correction based on the evidence. If patients cannot identify the intended practice, revise that identification under the approved message process. If they cannot locate the form after login, test the navigation or task label. If the form rejects a valid entry, investigate the configured validation.

Keep the task, eligibility rules and observation window stable while comparing results where practical. Record any changes in the invited population or account starting state. Comparing existing users with first-time enrollees can make a redesign appear more effective without demonstrating an easier journey.

Use three views of the result: portal task completion, completion through any supported route, and staff assistance effort. If portal use rises while total completion stays flat and staff work increases, the project has changed the channel mix. It has not yet demonstrated less work or better completion overall.

The healthcare call center workflow guide explains why a contact and an underlying request are different units. Apply that distinction when a patient calls about the same form they attempted online. Do not count the call as a second administrative need.

Retest the exact failure with synthetic data before promoting the changed journey broadly. Then review the next cohort's evidence. A clearer invitation is useful only if the subsequent task remains understandable and supported.

Keep the promise narrow as adoption expands

After one task works, decide whether the next task shares the same access route, required information, receiving owner and completion evidence. Reuse the proven elements and test what changes.

Do not infer that a patient who submitted a contact form can navigate every portal feature. A new task may have different permissions, unfamiliar terminology or a different destination team. The patient intake software guide addresses broader intake capabilities; this adoption method verifies how a particular patient journey works in practice.

Keep task-specific instructions current when menus, forms or supported routes change. Assign an owner to the patient-facing promise as well as the technical form. Someone should be able to verify that what the invitation says still matches what a patient encounters.

For the phone side of patient access, explore AI voice agents for healthcare.

FAQ

Sources

• ONC: Individuals' Access and Use of Patient Portals and Smartphone Health Apps, 2024, Data Brief 77, July 2025. National HINTS-based access context, distinct from local task completion.
• ONC Patient Engagement Playbook, enrollment assistance and clear instructions (chapter 2) and proxy access for caregivers (chapter 5).