Bright Spot Analysis — Find What Works and Test Why

Where does the work go better, and what is different there? Investigate promising examples, understand the conditions behind them and test what could help elsewhere. Copy the mechanism, not just the visible activity.

This method supports Right Problem? by revealing possible explanations, Right Intervention? by suggesting changes to test, and Right Measures? by checking whether those changes help. Move between the questions as you learn.

What is a Bright Spot?

A Bright Spot is a team, service or organisation achieving a better outcome under broadly comparable conditions. It is a starting point for investigation, not a label that proves excellence or identifies a cause.

The approach comes from positive deviance: look for people finding useful ways forward within shared constraints. A promising example need not be perfect. Investigate the particular outcome that is better, rather than treating the whole organisation as a model.

A high ranking or one good month may reflect ordinary variation, different resources or a different population. A persistent difference or a change in a time series strengthens the case for investigation, but neither proves that a practice caused it. A service can also perform consistently well without a recent change point.

How to investigate one

1. Define the outcome and check the comparison. What is better, for whom and over what period? Check definitions, missing data, demand, case mix, resources and possible harm elsewhere. Use suitable time-series or comparative analysis; do not rely on a league table alone.
2. Observe how the work happens. Follow real cases, talk to staff and service users, and examine decisions, delays and handovers. Ask what makes the work possible, not only what people say they do differently.
3. Build and challenge an explanation. Identify a practice or condition that could plausibly affect the outcome. Check when it began, what should change first and what other explanations fit. Choose an investigation window suited to the expected effects, not a fixed number of months.
4. Separate the mechanism from its local conditions. Which capabilities, authority, staffing, relationships or external services support it? Some conditions can be developed elsewhere; others may make transfer impractical. Record uncertainties rather than assuming the practice will travel.

For example, a daily huddle may help because staff can raise concerns safely and someone present can resolve them. Copying the meeting time without those conditions copies the appearance, not the mechanism.

How to test replication

Treat the proposed explanation as a hypothesis. Agree one bounded test with the people doing the work, appropriate authority, an owner and a review date. Check dependencies and safety before starting; high-risk or hard-to-reverse changes require proportionate investigation and oversight.

• Predict: state what should change, in which direction and by when, before seeing the result.
• Measure: check whether the practice was used, whether the expected early response occurred, whether the outcome improved and whether harm, delay or workload moved elsewhere.
• Review: allow for variation and realistic delays, then continue, adapt, stop or expand the test. Check persistence as conditions and leadership change.

Run charts, SPC and Bootstrap CUSUM can support evaluation where appropriate to the data and question. A detected change point is evidence of a change in the measured series, not proof of its cause or permanence. No detected change does not automatically mean no benefit. See Variation and SPC and Why Bootstrap CUSUM? for the methods.

Do not scale merely because one site succeeds. Test whether the explanation and supporting conditions hold in another setting. The PDSA cycle provides a practical learning structure.

Examples to explore

• Corridor Care Bright Spots: five case studies exploring whole-hospital coordination, discharge and diagnostic flow, with different kinds of evidence.
• Stowhealth — GP access: demand-matched staffing, clinical contact and team roles.
• Water and Sewage Bright Spots: examples for examining how operating conditions, investment and governance differ.

Use these as questions to investigate, not recipes or equivalent demonstrations of causal effectiveness.

Choose your next step

Bring one promising example and one uncertainty. Return to the question that would help most:

• Right Problem? — what difference might explain the better result?
• Right Intervention? — what could we responsibly test here?
• Right Measures? — what evidence would change our decision?