Make Capacity Trade-Offs Visible Without Manufacturing Precision
Estimate how executing units currently distribute capacity. The goal is not a perfect resource model; it is an honest conversation about constraints, fixed demand, flex capacity, and the realism of the emerging Value Stream Roadmap.
- Fast exercise
- 5 minutes
- Full exercise
- 15–25 minutes
- Input
- Units + objectives + initiatives
- Output
- Capacity forecast + confidence
- Next stage
- Flow Problems
Understanding Value Stream Capacity
Estimate only the detail that improves a decision
Capacity is a system-level constraint, not a promise that every person can be allocated at 100 percent. Use rough unit-level distributions to expose trade-offs, persistent demand, scarce capability, and the room available for new shared initiatives.
Understand social scale and rough executing capacity.
Name the trade-offs the value stream needs to see.
Separate historical data, credible estimate, and unknown.
Create a rough distribution only where it changes a decision.
More buckets and decimal places do not create better capacity insight. They may simply hide disagreement behind arithmetic.
Running the Exercise
Five moves from rough scale to realistic trade-offs
-
1
Approximately 1–3 minutes
Count the Executing Units
Bring forward the Value Stream Landscape. For each unit that performs material work, record rough people and team counts. Keep enabling and external capacity visible when it can constrain delivery.
ART / STPersistent delivery capacity Teams and trains executing features, capabilities, and large-solution work.
Domain / SACloser collaboration capacity Solution Areas, components, or domains that carry shared work.
PlatformShared enabling capacity Infrastructure, toolchain, data, DevOps, test, or AI services.
SupplierContracted or constrained capacity External teams, services, certification, labs, or partner delivery.
FunctionScarce expertise and authority Security, compliance, architecture, legal, procurement, or operations.
Use ranges when a precise count adds no value. The purpose is to understand scale and constraint—not to turn people into interchangeable units.
Speaker Notes
Facilitator intentEstablish the unit-level capacity boundary from the Value Stream Landscape: count people and teams roughly, keep enabling and external constraints visible, and avoid treating headcount as throughput or as interchangeable capacity.
How to facilitate
- Bring the Value Stream Landscape forward and create one row for every unit that performs material work. Use a rough people or team count—or a range—rather than pausing for exact HR data.
- Distinguish persistent delivery units from shared platforms, suppliers, labs, and scarce central expertise. Note whether their capacity is dedicated, shared, contracted, or available only in specific windows.
- State the planning horizon and a simple double-counting rule. Matrixed people and shared specialists should not quietly appear as fully available capacity in several units at once.
Listen and watch for
- Headcount being described as free capacity even though operations, support, maintenance, leave, and committed work already consume part of it.
- Supplier team size being confused with capacity the value stream can actually direct, or the same architect, test expert, or environment being counted in several places.
- Material contributors being omitted because they are not organized as an ART—for example a platform, test lab, operations group, certification body, or traditional department.
Avoid
- Turning the first step into individual utilization planning. The exercise needs a system-level view for trade-offs, not a person-by-person allocation sheet.
- Stopping for organizational data cleanup. Record a range, confidence, and validation owner, then continue up the depth ladder.
- Converting people counts directly into delivery forecasts before the work mix, evidence source, and constraints are visible.
-
2
Approximately 2–4 minutes
Choose Useful Capacity Buckets
Start with a recognizable set, then merge, rename, or remove buckets. Retain only categories that make a meaningful trade-off visible.
01Small enough Few enough to estimate quickly and consistently.
02Different enough Each represents a meaningfully different trade-off.
03Measurable enough Someone can estimate or observe it credibly.
04Decision-relevant It can change a roadmap, priority, or capacity decision.
New FeaturesEnablersMaintenanceKeep the Lights On Compliance / SafetyInnovation / LearningUnplanned BufferOperationsWhich buckets would help this value stream see whether the overall Value Stream Roadmap is realistic?
Speaker Notes
Facilitator intentChoose the smallest understandable set of work-type buckets that exposes meaningful roadmap trade-offs. The buckets are a planning lens for this value stream—not a universal taxonomy or a mandated percentage model.
How to facilitate
- Start with the deck's common set—New Features, Enablers, Maintenance, KTLO, Compliance or Safety, Innovation or Learning, and Buffer—then merge, rename, or remove categories for the local context.
- Give each retained bucket a one-sentence inclusion rule. Clarify recurring ambiguity, especially Maintenance versus KTLO, Compliance versus ordinary feature work, and Enablers versus general technical activity.
- Test every bucket against the four source criteria: small enough, different enough, measurable enough, and decision-relevant enough to change a roadmap or prioritization conversation.
Listen and watch for
- Overlapping buckets that allow the same work to be counted twice—for example a regulatory defect classified as Maintenance, Compliance, and KTLO.
- Different units using the same label for different work, or different labels for the same trade-off, which makes cross-unit comparison misleading.
- Unplanned Buffer being treated as a hidden wish list or as guaranteed spare capacity rather than protection against uncertainty and unplanned demand.
Avoid
- Letting the exercise become a taxonomy hobby. If a distinction cannot influence a decision, merge it.
- Treating sample values or external capacity-allocation examples as required targets. The deck deliberately asks the value stream to choose useful buckets for its own decisions.
- Adding a category that nobody can estimate, explain, or use. Keep a comparable core set and add local detail only where it earns its place.
-
3
Approximately 2–4 minutes
Name the Source and Confidence
Before discussing percentages, identify where the estimate comes from. Distinguish observed history from a leader estimate and from a planning assumption.
ObservedHistorical work data Completed-item history, incident load, maintenance demand, or actual allocation records.
EstimatedCredible role estimate RTE, PM, delivery, supplier, platform, or operations view informed by recent work.
AssumedPlanning hypothesis A provisional distribution to be validated before it drives a material commitment.
UnknownVisible evidence gap No credible source yet; assign an owner and date rather than inventing a number.
H · corroborated data M · credible estimate L · planning assumption ? · no source yetOwn idea: Record a “data freshness” date. A capacity estimate can be well sourced and still be stale after a reorganization, incident, supplier change, or regulatory escalation.
Speaker Notes
Facilitator intentMake the evidence behind every capacity claim visible before percentages gain authority: distinguish observed history, credible estimates, planning assumptions, and unknowns, including how fresh and representative the evidence is.
How to facilitate
- Ask four things for each estimate: source, accountable interpreter, time period, and data-freshness date. A well-known source can still be stale or unrepresentative.
- Triangulate where practical: compare work-system data with the informed view of RTEs, Product Management, platform or operations leads, suppliers, and finance rather than forcing one source to tell the whole story.
- Mark confidence H, M, L, or unknown and assign a validation owner and trigger when low-confidence data could materially change a roadmap decision.
Listen and watch for
- Tool reports whose categories reflect ticket-label habits rather than the work actually performed, especially for support, maintenance, enablers, and invisible coordination work.
- An exceptional period—a major incident, release, reorganization, supplier disruption, or regulatory escalation—being presented as the normal capacity mix without qualification.
- Numbers shaped by aspiration or budget defense: the desired future allocation is quietly substituted for the current or historically observed one.
Avoid
- Confusing precision with accuracy. A corroborated range can be more decision-useful than a detailed percentage produced by one weak source.
- Silently converting unknown capacity into zero, full availability, or the preferred answer. Unknown is a legitimate and actionable preparation gap.
- Waiting for a perfect reporting system. Stop at source and owner when necessary, then continue with explicit confidence.
-
4
Approximately 3–8 minutes
Estimate a Rough 100% Split
Use one row per executing unit. The values should roughly add to 100 percent. If they do not, the conversation has exposed a hidden demand or inconsistent boundary.
Capacity by Executing Unit Unit People New Enablers Maintenance KTLO Compliance Innovation Buffer Source / confidence ART A 100 35% 15% 15% 20% 5% 5% 5% RTE estimate / M ART B 85 28% 18% 12% 22% 8% 6% 6% Jira + PM / M Supplier X 40 22% 20% 10% 18% 20% 4% 6% supplier PM / L Platform 60 32% 12% 18% 20% 8% 4% 6% lead estimate / M ART A Supplier X Replace all sample values with participant data. If time is short, fill only people, the two largest buckets, source, and unknowns.
Speaker Notes
Facilitator intentCreate a rough, transparent 100-percent distribution for each executing unit so hidden demand, double counting, and incompatible assumptions become discussable; use the sample rows only to calibrate format, never as targets.
How to facilitate
- Use one row per executing unit and round numbers or ranges. Label whether the row describes recent actuals, the current expected mix, or a future planning hypothesis—do not blend these silently.
- Check that every row roughly adds to 100 percent. Treat a mismatch as useful evidence of hidden work, overlapping buckets, an inconsistent unit boundary, or an unspoken availability assumption.
- Use the deck's fast mode when time is short: capture people, the two largest buckets, source and confidence, plus unknowns. Deeper precision can become preparation work.
Listen and watch for
- Totals above 100 percent because committed initiatives, maintenance, compliance, and support have each been estimated independently without one common capacity boundary.
- Shared specialists, platforms, environments, or supplier capacity appearing as fully available in several rows at once.
- A desired allocation being presented as current reality, or the deck's ART A, ART B, Supplier X, and Platform sample values being treated as benchmarks.
Avoid
- Averaging the units too early. The value of the canvas lies in seeing where their mixes and constraints differ.
- Using the percentages to allocate every individual or to imply that all working time should be filled. This is a unit-level trade-off view, not a utilization target.
- Defending the arithmetic when the row feels wrong. Discuss the boundary, source, or hidden assumption instead.
-
5
Approximately 3–5 minutes
Read Constraints, Fixed Demand, and Flex Capacity
Do not stop at percentages. Compare objective and initiative demand with the units that must absorb it. Identify where a trade-off, sequencing decision, or capacity conversation is unavoidable.
Persistent demandWhat already consumes capacity? Operations, maintenance, compliance, support, and committed work.
Shared demandWhat crosses units? Objectives and initiatives requiring several units or scarce shared services.
ConstraintWhere is the narrowest capacity? Specialist, supplier, platform, environment, approval, or evidence bottleneck.
FlexWhat can actually move? Buffer, timing, sequence, scope, sourcing, or investment—not imaginary spare people.
Decision-ready observation [Demand] competes for [constrained capacity] in [units], leaving [credible flex] and requiring [trade-off or sequencing decision].Example: “Three shared initiatives depend on the same test platform while compliance already consumes 20% of Supplier X capacity; the roadmap needs an explicit integration-window decision.”
The exercise succeeds when the group can have a more honest conversation about capacity than before.
Speaker Notes
Facilitator intentTurn the capacity canvas into decision-ready observations: connect objectives and shared initiatives to persistent demand, constrained capabilities, credible flex, and the work the value stream cannot absorb without an explicit trade-off.
How to facilitate
- Overlay the Business Objectives and Value Stream Initiatives on the capacity rows. Ask which units and shared services must absorb each new or cross-unit demand.
- Identify the narrowest relevant capacity, not merely the smallest headcount: scarce skill, supplier window, platform, environment, approval, evidence chain, or integration capacity may govern the outcome.
- Write one decision-ready observation per material constraint: demand, constrained capacity, affected units, credible flex, and the trade-off or sequencing decision required.
Listen and watch for
- Apparent spare capacity that has the wrong skill, timing, location, contractual freedom, or dependency context to help the constrained work.
- Buffer that is already consumed by normal variability, incidents, or unplanned obligations, and therefore is not credible flex for another initiative.
- Every bucket being described as fixed. That is a signal to test feasibility or escalate the roadmap—not a reason to hide the gap in Quality, overtime, or optimistic assumptions.
Avoid
- Treating the rough canvas as a commitment. It is a realism filter and a source of questions for conference preparation and later refinement.
- Assuming capacity can be solved by moving people between stable teams. First examine priority, sequence, scope, sourcing, investment, and the Flow of work through the constraint.
- Solving every Flow problem now. Capture the capacity-related decision backlog and hand it to the next stage, where Flow Problems are diagnosed and prioritized.
Facilitation
Keep capacity directional, comparable, and decision-relevant
Ask which distinction could change a roadmap or trade-off. Merge the rest.
Use ranges, round numbers, and confidence. A credible 20–30% is better than an invented 24.7%.
Return to unit-level Flow. Leave room for collaboration, learning, variability, and sustainable work.
Celebrate the signal. The group has found hidden demand, double counting, or incompatible assumptions.
Ask who fixed it, for what period, with what consequence, and which lever remains open.
Stop at sources and owners. Do not let a missing report block the rest of the diagnostic path.
Ready to Move On?
The capacity view is ready when it changes the realism of the conversation
- Every material executing unit has a rough people or team count.
- The bucket set is small, distinct, measurable, and decision-relevant.
- Persistent demand such as maintenance, KTLO, compliance, and operations is visible.
- Each material estimate has a source and confidence level.
- Data freshness or major recent changes are noted.
- Rows roughly add to 100 percent or expose a named inconsistency.
- Shared objectives and initiatives are compared with available unit capacity.
- Scarce capability, supplier, platform, or approval constraints are visible.
- Credible flex and real trade-off levers are named.
- Material evidence gaps have owners and follow-up dates.
Source Foundation
This page distills slides 14–17 of 2026-07-13_Value_Stream_Conference.pptx. Current SAFe portfolio guidance supports visualizing and limiting WIP, balancing new development with ongoing maintenance, and treating capacity and investment as value-stream concerns.