You Know What Hurts. Now Find Where the Flow Breaks.
Start with the ranked challenge views from Exercise 2. Trace the observable value-stream mechanisms behind them, keep every plausible link visible, and build one prioritized Flow-problem backlog per group.
- Core exercise
- 20 minutes
- Total section
- Approximately 25 minutes
- Format
- Think → Map & Rank → Share
- Group size
- 3–4 participants
- Output
- One linked, ranked Flow-problem backlog per group
Understanding Flow Diagnosis
Move one level deeper before discussing solutions
A missed date, recurring defect, or slow approval describes what people experience. Flow Diagnosis asks what the current work system does that may create or amplify that experience. The result is a set of causal hypotheses—not a verdict and not yet a countermeasure.
Keep the ranked challenge views, the validated constraints, and the protected Quality boundaries visible throughout the exercise.
What people experience: a missed commitment, delay, defect, overload, or unreliable outcome.
Where work waits, piles up, loops back, crosses boundaries, or repeatedly loses focus.
What the organization may change after evidence has strengthened or refuted the hypothesis.
Why Mechanisms Come First
Diagnose the current system before designing its future
Jumping directly from a symptom to a preferred solution hides the causal assumption. A useful Flow problem can be located in the current value stream and expressed in observable terms: what is waiting, where it waits, what creates the queue or loop, and which challenges may result.
Map what actually happens—including material, information, technical, and decision Flow—before selecting an intervention.
Eight Diagnostic Lenses
Look for where work stops, piles up, loops, or loses focus
Use these prompts as search lenses, not as an official or exhaustive taxonomy. The final card should be more concrete than the lens that helped the group see it.
More work starts than finishes.
One step cannot keep up with demand.
Work or information waits across boundaries.
Ready work waits for attention or authority.
Feedback arrives only after too much work.
Work loops back after assumptions or defects surface.
Parallel demand prevents focused completion.
Approvals, rules, or cadences create systematic waiting.
Three Mapping Patterns
The mapping is rarely one-to-one
Every connection is a hypothesis. Preserve all plausible links so the map can expose shared mechanisms and compound causes.
A single visible challenge has one leading mechanism worth testing first.
One system mechanism may generate different symptoms across the value stream.
Several mechanisms may combine to produce one repeatedly visible outcome.
A mechanism connected to several high-priority challenges may offer broad leverage; a challenge with several incoming mechanisms warns against a single-cause story.
Running Flow Diagnosis
Nine moves from ranked challenges to prioritized Flow hypotheses
The conceptual frame is compressed to approximately five minutes. The exercise then uses an exact twenty-minute Think → Map & Rank → Share rhythm.
-
1
Approximately 20–30 seconds
Open the Diagnosis
Leave the ranked challenge views from Exercise 2 fully visible. Explain that the groups will not solve those challenges yet; they will look for the mechanisms that may create or intensify them.
You already know what hurts. We will now look for the places and mechanisms in the value stream that may be producing these challenges.
InputRanked challenge views from Exercise 2 OutputOne linked, ranked Flow-problem backlog per group Speaker Notes
Facilitator intentOpen a bounded diagnosis phase: keep the ranked challenge views, validated constraints, and protected Quality boundaries visible while the group moves from experienced effects to possible Flow mechanisms without solving them yet.
How to facilitate
- Re-anchor each group in its own ranked challenge view from Exercise 2. Keep the separate views intact rather than merging them into one master list.
- State the input, output, and guiding question in plain language: which mechanisms in the current value stream may create or amplify these challenges?
- If the discussion becomes broad, name the Flow item and system boundary the group will follow—for example a feature, interface decision, evidence package, release, or customer request.
Listen and watch for
- A tool, reorganization, new meeting, or preferred method being presented immediately as the diagnosis.
- Participants merely rewriting a challenge card instead of locating where work waits, piles up, loops back, or loses focus.
- Hierarchy or a confident first speaker replacing the evidence and ranking the group already produced in Exercise 2.
Avoid
- Starting a new cross-group prioritization round; the groups need their own evidence trail for this exercise.
- Removing the original challenge cards. The visible links between experienced effects and proposed mechanisms are part of the output.
- Presenting the coming map as a proven causal model. It is a transparent set of hypotheses to validate.
-
2
Approximately 30 seconds
Separate Symptom, Mechanism, and Countermeasure
Make the three analysis levels explicit. Today the group identifies and ranks the mechanism; it does not design the intervention.
Do not jump from the symptom straight to your preferred solution. First describe how the current Flow may create the symptom.
Speaker Notes
Facilitator intentProtect the distinction between an experienced challenge, a Flow mechanism in the current system, and a future countermeasure so the room does not confuse a preferred action with an explanation.
How to facilitate
- Test every statement with one question: is this the observed effect, the mechanism that may create it, or already a countermeasure?
- Make the mechanism observable by naming the work item, the location or state in the value stream, and what waits, accumulates, returns, or repeatedly loses focus.
- Park proposed interventions in a clearly separate area. Acknowledge them without allowing them to replace the current-state diagnosis.
Listen and watch for
- Abstract nouns such as culture, governance, architecture, alignment, or communication without a described Flow mechanism.
- Statements about a person's or team's motivation, competence, or speed instead of the policy, queue, dependency, feedback loop, or capacity constraint around the work.
- Action language—implement, create, introduce, reorganize, automate—that signals the group has already moved into solution design.
Avoid
- Using the 5 Whys or another causal technique as proof. It helps generate better hypotheses; it does not establish causality by itself.
- Debating whether a countermeasure is good before the mechanism it is supposed to change has been made explicit.
- Forcing a single root-cause story in a complex system. Several mechanisms may contribute at the same time.
-
3
Approximately 30 seconds
Keep Every Plausible Link Visible
Introduce the many-to-many logic before anyone begins writing cards. A challenge may have several contributing mechanisms, and one mechanism may help explain several challenges.
Keep every plausible causal link visible. We are building a hypothesis map, not forcing a tidy one-to-one classification.
Speaker Notes
Facilitator intentPreserve causal complexity long enough to see shared mechanisms and compound causes: the arrows are hypotheses, not causal facts, and many-to-many mapping is expected.
How to facilitate
- Draw every plausible link using provisional language such as may contribute to, may amplify, or is consistent with.
- Allow one Flow problem to connect to several challenges and one challenge to receive links from several Flow problems.
- Consolidate cards only after comparing both location and mechanism. When two cards are merged, preserve every incoming and outgoing link.
Listen and watch for
- A desire to assign each challenge to exactly one category because the resulting picture looks cleaner.
- The number of links being treated as evidence that a hypothesis is true rather than as a signal of possible reach.
- Two cards with the same label but different locations, work items, or mechanisms being collapsed too early.
Avoid
- Tidying the map for presentation aesthetics at the cost of losing uncertainty or minority interpretations.
- Deleting a plausible link merely because most participants prefer another explanation.
- Describing the arrows as established causes. The map makes assumptions inspectable; validation comes later.
-
4
Approximately 1 minute
Scan the Eight Diagnostic Lenses
Offer the lenses as prompts. Then ask participants to make every Flow-problem card specific to a location, a work item, and an observable mechanism.
Too abstract: “Too much WIP.” Observable: “Five product initiatives are simultaneously waiting for the same integration team.”
Speaker Notes
Facilitator intentUse the eight lenses as search prompts—not answer categories—then translate any useful prompt into a concrete, observable Flow-problem statement.
How to facilitate
- Scan the first four lenses around starts, constraints, boundaries, and queues; then scan the second four around batch size, feedback, focus, and decision policies.
- For every candidate, ask for three specifics: which work item, where in the value stream, and what observable delay or loop occurs there?
- Ask which elapsed state or Flow signal could reveal the condition—for example WIP, Work Item Age, Queue Time, Blocked Time, rework, or decision wait.
Listen and watch for
- Cards that repeat the lens itself—too much WIP, bottleneck, dependency—without locating the condition in real work.
- A Flow accelerator or proposed remedy being written where the current inhibitor should be described.
- Local utilization being mistaken for end-to-end Flow: a busy unit may still create a longer queue for the whole system.
Avoid
- Requiring the group to use all eight lenses or to produce one card per lens. They are prompts, not a completeness checklist.
- Presenting the list as an official, exhaustive Flow-problem taxonomy. The deck deliberately uses it as a diagnostic aid.
- Measuring effort alone. The diagnosis is especially interested in elapsed time and the states in which work is not advancing.
-
5
Approximately 1½ minutes
Practice the Three Mapping Patterns
Use the examples to show the level of specificity required. They are intentionally simple illustrations, not external case data.
1 ↔ 1 Regulatory approval arrives too late Evidence: completed dossiers wait for a monthly board.
Hypothesis: a decision queue and cadence create waiting.
N → 1 Late starts + late integration defects Shared mechanism: interfaces are discovered only after work starts.
1 → N Launch date repeatedly missed Mechanism A: high WIP slows completion.
Mechanism B: late integration creates rework.
What in each example can we actually locate and observe in the value stream?
Speaker Notes
Facilitator intentCalibrate the room on the required specificity and on the three mapping patterns, while making clear that the examples are illustrative calibration—not external case evidence or content to copy.
How to facilitate
- For each example, ask the room to identify the observed challenge, the located mechanism, the available evidence, and at least one alternative explanation that is still possible.
- Contrast the patterns explicitly: one leading mechanism, one shared mechanism behind several effects, and several mechanisms contributing to one effect.
- Close each example with the same question: what can we actually locate and observe in the current value stream?
Listen and watch for
- Participants copying regulatory boards, unclear interfaces, or launch dates into their own map because the examples are vivid.
- Evidence and hypothesis being blended—for example, treating a monthly cadence as proof that it caused the missed date.
- Confidence rising simply because an example has a neat one-to-one shape. Simplicity does not make a causal claim more certain.
Avoid
- Spending so long on the examples that participants have less time to work with their own evidence.
- Introducing the solution to the example. The teaching point is the structure of the diagnosis, not the countermeasure.
- Implying that one mapping pattern is superior. The useful pattern is the one that preserves the plausible relationships in the case.
-
6
Approximately 45 seconds
Write Testable Flow Hypotheses
Each card names a location and mechanism, links to every challenge it may explain, and identifies evidence that could strengthen or refute the claim.
Sentence frame We believe [mechanism] at [location] contributes to [linked challenge(s)], as indicated by [observation or data].01Location Where in the value stream?
02Mechanism What waits, piles up, loops, or loses focus?
03Links Which challenge cards may it explain?
04Evidence What would confirm or refute it?
Translate personal blame into system language. “Team X is too slow” is not a Flow mechanism. If evidence is missing, a question-shaped hypothesis is acceptable.
Speaker Notes
Facilitator intentTurn each Flow-problem card into a claim that can become more or less credible through observation, including a clear answer to the question: what would change your mind?
How to facilitate
- Use the full sentence frame: mechanism, location, linked challenge or challenges, and the observation or data that currently supports the claim.
- Distinguish evidence already available from evidence the group would need to collect after the workshop.
- Ask explicitly what result would weaken or refute the hypothesis, not only what would confirm it.
Listen and watch for
- Claims so broad or elastic that every possible observation could be interpreted as confirmation.
- Unsupported numbers or precise causal language that create an appearance of certainty the group does not yet have.
- Blame, intent, or competence language. Translate it into an observable policy, queue, handoff, interruption, feedback delay, or capacity condition.
Avoid
- Demanding complete datasets during the twenty-minute exercise. Naming the missing evidence is already useful progress.
- Treating correlation, sequence, or a compelling story as causal proof.
- Embedding the countermeasure in the hypothesis—for example, we need a weekly board—rather than describing the current mechanism.
-
7
Approximately 30 seconds
Explain the Twenty-Minute Rhythm
Keep the same groups and ranked challenge views from Exercise 2. Make the complete sequence and expected output visible before starting the first timer.
Think3 min Propose Flow-problem hypotheses.
Map & Rank12 min Link, consolidate, and rank.
Share5 min Show the highest-leverage Flow problems.
Outcome: one ranked Flow-problem backlog per group, linked to Exercise 2.The groups remain separate. Do not create a cross-group master ranking during this exercise.
Speaker Notes
Facilitator intentMake the complete Think → Map & Rank → Share sequence, its exact timing, and its expected artifact visible before anyone begins, so all participants hear the whole sequence before the first timer starts.
How to facilitate
- Keep the same groups and ranked challenge views from Exercise 2, then walk through the 3-minute, 12-minute, and 5-minute phases without starting early.
- Show what one finished card, its many-to-many links, and the relative ranked backlog should look like.
- State the boundary clearly: each group produces its own complete backlog; there is no cross-group master ranking in this exercise.
Listen and watch for
- Questions about solutions, action owners, or implementation plans that belong after diagnosis and validation.
- One group beginning to discuss before quieter participants or remote participants have heard the full instruction.
- Hybrid participants lacking access to the board, cards, links, or timer before the exercise starts.
Avoid
- Adding new instructions during the silent Think phase because the initial explanation was incomplete.
- Allowing the timeboxes to become vague suggestions; the rhythm protects both independent thought and group synthesis.
- Changing group composition without a real need. Continuity preserves the meaning of each group's Exercise 2 ranking.
-
8
3 quiet minutes
Think Individually
Protect independent diagnosis before the group begins interpreting the same challenge views together.
01Start with the Ranking Use your group’s challenge view from Exercise 2.
02Ask Where Flow Breaks Where does work wait, pile up, loop, or cross a boundary?
03One Hypothesis per Card Write up to three possible Flow problems.
04Keep the Links Mark every challenge card the hypothesis may explain.
Take three quiet minutes. Write up to three possible Flow problems, one mechanism per card, and mark every challenge each card may help explain.
Start the timer and stop speaking. No solution or action cards yet.
Speaker Notes
Facilitator intentProtect three minutes of independent search before discussion to broaden the hypothesis pool and reduce production blocking and collaborative fixation on the first explanation voiced in the group.
How to facilitate
- Ask for up to three cards per person, one possible mechanism per card, with every linked challenge visibly marked.
- Leave the eight lenses available as quiet prompts, but do not seed the room with additional examples once the timer begins.
- Start the three-minute timer and stop speaking. Let participants work independently before any interpretation is shared.
Listen and watch for
- Participants waiting to see what a senior leader or recognized expert writes before forming their own view.
- Cards that simply restate a challenge or already prescribe a countermeasure.
- Side conversations that remove the independence of the phase or disadvantage remote participants.
Avoid
- Circulating through the room and coaching the content of individual hypotheses; answer process questions only.
- Forcing certainty. A question-shaped hypothesis is acceptable when the participant can name what should be investigated.
- Ending the silence early because some people have finished. The remaining time gives others space to think beyond the obvious first idea.
Ready to Move On?
The group is ready for the Value Stream Conference when …
Use this as a completion check. Do not turn it into a hidden cross-group re-prioritization round.
- The ranked challenge views from Exercise 2 remain visible.
- Every Flow-problem card names an observable location and mechanism.
- Problems, Flow problems, and countermeasures are clearly separated.
- Every plausible challenge-to-mechanism link remains visible.
- Duplicate mechanisms are consolidated without losing their links.
- Each group has one complete relative Flow-problem ranking.
- Expected leverage considers relevance, reach, and plausibility.
- Each group explained its top three and stated the expected effect.
- Evidence that could confirm or refute each top hypothesis is named.
- The group backlogs remain separate, provisional, and ready for a decision-capable forum.
The diagnosis converts daily friction into a transparent hypothesis backlog without pretending that the causes are already proven or that the solution is already known.
Facilitation Notes Keep causal language provisional and system-focused Open preparation guidance
Facilitation
Keep causal language provisional and system-focused
The facilitator protects the boundary between observation, explanation, and intervention. Good diagnosis is specific enough to test and humble enough to be wrong.
Facilitation principles
- Keep all ranked challenge views visible.
- Ask for a location and observable mechanism.
- Allow many-to-many links.
- Consolidate duplicate mechanisms without losing connections.
- Translate blame into system language.
- Keep countermeasures out of the diagnosis.
- Rank by expected leverage, not certainty.
- Keep group backlogs side by side.
Useful interventions
What can we actually locate and observe in the current value stream?
Is that the experienced problem, the Flow mechanism, or already a countermeasure?
Which other challenge might this mechanism help explain?
What observation or data would make this hypothesis more or less credible?
Let us describe the policy, queue, handoff, or feedback loop rather than blaming a person or team.
Validate After the Workshop
Match each hypothesis to evidence that can change your mind
The twenty-minute exercise identifies where to investigate. It does not prove causality. Start with the minimum Flow metrics and add context-specific signals where the hypothesis requires them.
How many work items are started but not finished?
How many work items finish per unit of time?
How long has current work already been in progress?
How much elapsed time passes from start to finish or release?
Where do Queue Time, Blocked Time, late feedback, and loops accumulate?
How often do handoffs occur, and how long do approvals or decisions wait?
Source foundation from the reference presentation
The web module distills the supplied 16-slide workshop deck Dynamic_Agility_Workshop_3_Mapping_and_Prioritizing_Flow_Problems.pptx. Its public reference base includes:
- SAFe: Accelerating Flow — Flow interruptions, Flow metrics, and system-level improvement
- SAFe: Solution Train Flow — identifying and addressing Flow impediments across large, complex systems
- The Kanban Guide — WIP control, active item management, and the minimum Flow metrics
- Lean Enterprise Institute: Value Stream Mapping — current-state material and information Flow and whole-system visibility
- Lean Enterprise Institute: 5 Whys — moving beyond obvious symptoms while separating analysis and countermeasures
- Lean Enterprise Institute: Seven Wastes — waiting, inventory, correction, and other observable waste patterns