Example (composite): How a community clinic used measurement, experiments, and huddles to cut waits

Note: This is an anonymized, composite example designed to illustrate how the measurement-to-experiment loop commonly plays out. It synthesizes patterns seen in multiple improvement efforts rather than reporting a single verified project.

Situation

A mid-sized outpatient clinic saw frequent patient complaints about long waits, unpredictable daily variation, and a rising LWBS rate. Leadership knew adding sessions was expensive but unclear where to invest.

What the team did

1. Baseline measurement: The team ran a 7-day measurement using EHR timestamps and manual spot checks to validate definitions. They captured median and p90 arrival-to-provider and LWBS.
2. Bottleneck mapping: Using a simple flow map and a few hours of observation, staff identified rooming delays caused by a single float nurse covering triage and rooming during peak hours.
3. Small tests: They ran two-week experiments: (a) assign a dedicated rooming nurse from 09:00–12:00 on high-volume days, and (b) pre-order common point-of-care tests for specific visit types so nurses didn’t need to leave the room mid-visit.
4. Daily huddle: A short morning huddle reviewed the dashboard (median, p90, LWBS), confirmed staffing assignments, and flagged any expected surges. Owners were assigned for any red items.
5. Scale & standardize: When data showed consistent improvements during test windows, the team converted the assignment into a standard work pattern and updated schedules.

Outcomes (illustrative)

In this composite example, the clinic saw steady reductions in median arrival-to-provider during the tested times and lower LWBS on the days the float nurse was assigned. The key point is not a single number but the pattern: measurement identified problem times, short tests produced quick learning, and huddles kept the team aligned.

Lessons learned

• Validate timestamps—automated data can hide mislabelled events.
• Small, staff-owned experiments are less disruptive and produce faster learning than large redesigns.
• Use the p90 as your early-warning signal—median can look fine while some patients still wait too long.
• Preserve gains by turning successful tests into simple standard work and including the change in onboarding for new staff.