Closing the loop on patient care

Juniper is Eucalyptus’ digital weight management clinic, treating obesity as a chronic condition through GLP-1 medications paired with coaching, clinician support, and ongoing care. Like any medication, GLP-1s come with side effects — nausea, fatigue, and GI symptoms among the most common — and how patients navigate those early weeks often determines whether they stay on treatment long enough to see results.

Internal data showed that 15–20% of cancellations were driven by side effect experiences and unmet support needs, with patients citing reasons like “I’m experiencing side effects” or “I didn’t get the support I needed.” Improving how patients log and act on side effects became a clear opportunity to reduce churn and strengthen the care loop between patient and clinician.

Parallel to the redesign of the side effect check-in flow itself, the Product team was restructuring the entire app IA. The existing IA and experience demonstrates:

• Lack of intuition and mental model for users to discover support features naturally.
• Support action is split under sections that indicate otherwise — like Progress and Chat — which is not intuitive at first glance.

The new mental model presents a new structure which includes the removal of Chat and the addition of Care — the new home for support. With clearer naming conventions, this enabled the many support actions to sit under one roof that is intuitive to find.

This blueprint maps the redesigned service in two sections:

• The User Journey — what the patient does.
• The System Logic — logic rules that trigger system operations underneath.

The new IA structure not only enables new check-in entry points (Home or Care) but it also supports the new check-in cadence changing from fortnightly to weekly. Once a patient logs their side effect and severity levels, the system decides what should happen rather than routing every report to a human. Together with the clinical team, severity routing logic was re-examined. Mild or Moderate logs lead to self-help tips and Severe routes to medical support.

However, severity alone doesn’t determine when human support is required — log frequency matters too. If the same side effect is logged twice or several times in a week, this can escalate a case even when no report is severe. These thresholds, set with doctors and coaches, gave the clinical team confidence that they could respond to every escalated case within the 24-hour window promised to patients. The new logic had to reflect not just clinical risk but also what the team could realistically act on.

The redesigned flow gets a patient from “I’m having a side effect” to useful guidance in two questions: what they’re experiencing, and how severe it is. Where the old check-in buried side effects behind weight and motivation questions, support is now its own flow — available any time from Care, so a patient can self-serve or reach a human whenever a side effect actually happens.

The key decision was what happens after those two questions. Rather than sending every report to a human, the flow forks on severity:

• Mild and moderate side effects resolve immediately into self-serve tips — practical, side-effect-specific guidance the patient can act on straight away. These cases don’t need to wait for a coach or clinician; making them wait was part of the old problem.
• Severe side effects route to the medical support team, with the patient told clearly that they’ll be contacted within 24 hours, plus the same tips as interim guidance.

Mild and moderate patients still get a route to a human, via a tertiary “Contact medical support” link. This was a deliberate V1 trade-off: customer support still runs on Zendesk, and rebuilding that whole intake was out of scope for this release. The goal here was narrower — make it easier for patients to log a side effect, and measure whether a shorter, self-serve-first flow works — so preserving the existing Zendesk path as a secondary option let us ship the core improvement without rebuilding support end-to-end.