Plain-language services. More booked appointments.

People were getting stuck before scheduling began

Company offered routine outpatient appointments across a region. Its public website asked visitors to choose a service category and location before opening the scheduler. The Director of Digital Access saw exits at that handoff but could not tell whether the issue was appointment availability, terminology or navigation.

The website asked visitors to know the organization

Daleel compared the public discovery steps in FullSession for the team. Heatmaps showed repeated clicks on service descriptions that looked actionable but were not links. Replays showed visitors moving between similarly named categories. An optional public-page feedback prompt revealed uncertainty about which label matched the service they wanted.

The UX Research Lead then tested the terminology with eight participants using sample appointment-search tasks. Six initially selected the wrong category for at least one task. The study explained the terminology problem; the experiment measured the change in completed appointments.

Simplify the route without collecting more sensitive detail

The team replaced internal department names with plain-language service labels, made each service card clickable and carried the selected location into scheduling. It added a clear route for people who needed help choosing. Capture was restricted to approved public pages: scheduling, symptom entry, patient identifiers and payment fields were excluded, and the exclusions were verified with synthetic sessions before the experiment.

Count completed appointments, not scheduler opens

In a six-week experiment, 9,967 eligible public-site visitors entered the control and 10,043 the treatment. Completed online appointments within 24 hours rose from 2,061 in control to 2,442 in treatment. The completion rate increased from 20.68% to 24.32%: 3.64 percentage points, or 17.6% relative improvement.

The scheduling team supplied aggregated counts by experiment arm from its own system. A limited experiment identifier connected assignment to completion without sending appointment details or patient identities into FullSession. Measured against the control rate, the treatment cohort produced about 365 additional completed online appointments.

How appointment completion was measured

Assignment was persistent on the first visit to an eligible service-discovery page. Both variants offered the same locations and appointment inventory during the same period. The denominator included visitors who did not open the scheduler, so the improvement is not an artifact of counting only those who made it through the first steps. Each entrant received the full 24-hour completion window. The comparison measures online completion, rather than total access across phone and online channels. Rates and relative improvement are rounded.

What the network retained

The Director of Digital Access approved the clearer navigation and kept service-card interaction and scheduler handoff in the review routine. Availability remained a separate constraint: clearer navigation could not create appointment slots. FullSession explained public-site behavior; the scheduling system remained the source of completed-appointment counts.