In this guide
- 01Agree the process boundary
- 02Check the current work
- 03Examine a problem
- 04Rehearse and commit
A process workshop should produce a shared account of current work, a defensible improvement question and an owned next test. The process map is a means to that result. A beautifully drawn flow with no agreed change or decision owner is unfinished work.
Use a workshop when several roles understand different parts of the same process and need to examine the connections. If the task is only to gather open concerns, a Lean Coffee may be a better starting format. If the proposed solution has already been imposed, clarify what participants can still influence.
Prepare the boundary, people and evidence
Choose a process with a recognizable trigger and result. For a hypothetical warehouse, the boundary might run from receiving a damaged-parcel report to giving the customer an agreed resolution. “Improve logistics” is too broad; “speed up customer service” may exclude the warehouse dependency that causes the delay.
Invite people who perform, receive and coordinate the work, together with someone who can authorize a next step. Include the perspective of the person using the service, through research or relevant participation. Stakeholder mapping can reveal missing roles whose work is affected even if they rarely attend improvement meetings.
Bring examples of actual case types and available operational records, with sensitive information removed. Ask participants to distinguish ordinary work, exceptions and stated procedure. Do not treat one memorable failure as evidence of its frequency.
NHS England’s process-improvement guidance combines process mapping, baseline measurement, testing and shared learning. This guide adapts those general disciplines to a non-clinical workshop. It does not reproduce a care process or establish that the same intervention works in a warehouse.
Map the work people actually do
Open with the task and the limits of the session. Ask participants to describe how a case moves, who acts, what information they use and where work waits. Draw the initial map with simple steps and role boundaries. Explain symbols as they appear; formal notation is unnecessary unless the team needs it.

In the fictional warehouse, customer service may wait for a photograph, inspection may wait for a parcel reference and the customer may believe a replacement is already on its way. These are possibilities to investigate, not claims about a real operation.
When accounts conflict, record both and identify how to check them. GOV.UK’s contextual research guidance describes observing tasks in their everyday setting. A targeted observation after the workshop may resolve a disputed step more reliably than another round of argument around the map.
Keep observation separate from interpretation. “The case stays in this queue until a supervisor returns” describes a proposed mechanism. “The other team does not care” assigns a motive that the map cannot establish. Use systemic questions to explore the conditions and interactions instead.
Give the session a realistic sequence
The following is an original agenda outline. Allocate time according to the process and participants rather than treating the outline as a standard duration.
| Part | Facilitation task | Output |
|---|---|---|
| Boundary | Agree the trigger, result and decision scope | A process sentence everyone can correct |
| Current work | Trace roles, information, waiting and exceptions | A provisional map with uncertainties |
| Problem examination | Select a difficulty and examine possible mechanisms | A question supported by available evidence |
| Improvement options | Develop and compare more than one response | A small set of plausible changes |
| Rehearsal | Walk through the proposed arrangement | Dependencies and exceptions exposed |
| Commitment | Agree the next test and review | Owner, support, measures and decision point |
Give quieter participants an opportunity to write or sketch before the discussion. Ask each role to correct its part of the map. If a manager’s presence makes disagreement difficult, provide another way to contribute and be honest about how the information will be used.
Investigate before choosing a fix
Select one problem that matters to the end-to-end result. Trace what may produce it and identify alternative explanations. A missing photograph might result from an unclear request, an inaccessible upload route, a customer who no longer has the packaging or an internal record that loses the attachment.
These explanations suggest different changes. Do not choose a reminder email before checking whether the information is obtainable or whether the organization can resolve the case without it. Customer-needs research and a service blueprint can extend the inquiry beyond internal assumptions.
Agree which uncertainty can be reduced now. Some questions can be answered by examining a few de-identified cases. Others require observation or a more deliberate research plan. Record what remains unknown rather than using consensus to turn a hypothesis into a fact.
Rehearse the future process before committing
Use a realistic case to walk through the proposed change. Ask each person what they receive, what they do, what they pass on and what happens when something is missing. Include an exception that challenges the design.

For the fictional warehouse, the group might rehearse a shared exception check before a case enters the inspection queue. The exercise could expose that a needed decision is outside the participants’ authority or that a proposed information field is not available at that point. These are useful findings to record before a live test.
The Institute for Healthcare Improvement’s testing guidance uses small, prediction-based tests to learn in a local setting. Applied here, a test plan should name the question, boundary, expected observation, support and next review. It should not turn a rehearsal into a claimed improvement result.
Use experiment design if the decision requires a stronger comparison. A workshop can design an inquiry; it cannot establish causality by agreement among participants.
Decide what to measure and who acts
IHI’s measurement guidance distinguishes outcome, process and balancing measures. In the warehouse example, resolution time might describe an outcome, completeness at intake a process feature and extra workload a possible balancing concern. Choose measures suited to the actual question rather than copying these examples automatically.
Agree definitions. When does resolution time begin and end? Does the record include unresolved cases? Who notices when work moves into an informal queue? A number can look better because the recording boundary changed. The reliability and validity guide explains why a consistent metric can still support the wrong conclusion.
Close with named responsibilities. The facilitator may document the agreement, but should not silently become the process owner. The person authorizing the test must understand its resource needs and conditions. Give participants a way to correct the record and communicate what happens after the session.
Failure modes and evidence limits
Common risks include mapping the ideal procedure, excluding the receiving role, voting on causes without evidence and approving a change that another department must deliver. A workshop can also mistake a shortage of capacity for a communication problem and create more meetings around the same queue.
The agenda and warehouse example are original practical adaptations. The sources support process inquiry, measurement and local testing; they do not validate this particular agenda or promise a productivity gain. Clinical-origin improvement guidance is used for general method concepts only.
Before the next session, write its intended output in one sentence: a checked map, a bounded problem and an authorized next test. If the group cannot name who can act on that result, resolve the mandate through change contracting before inviting everyone to another workshop.
Sources:
- NHS England (2024). How to improve care related processes in general practice.
- Government Digital Service (n.d.). Contextual research and observation.
- Institute for Healthcare Improvement (n.d.). Model for Improvement: Testing Changes.
- Institute for Healthcare Improvement (n.d.). Model for Improvement: Establishing Measures.



