The capacity model decay pattern
You spent two weeks building the capacity model. Product and engineering finally agreed on effective capacity. Discovery kanban shrank uncertainty from 4× to 1.1×. You walked execs through committed/target/stretch scenarios and got real buy-in.
Then sprint planning happened. Jira tickets piled up. Someone left, someone joined. The discovery board fell behind. Six weeks later, your capacity model is a stale artifact and everyone's back to "let's just commit and see what happens."
"We built this amazing capacity model in Q2. Presented scenarios to execs, got alignment, felt like we'd finally cracked roadmap planning. By August nobody was updating it. September roadmap review was back to gut-feel estimates."
This is the capacity model decay pattern, and it follows a predictable timeline:
- Week 0-2: High engagement, model feels revolutionary, team religiously updates discovery board and capacity inputs
- Week 3-6: Updates slow down as sprint pressure mounts, "we're too busy shipping to maintain planning artifacts"
- Week 7+: System effectively dead, back to old habits—JIRA-only workflow, wishful thinking estimates, quarterly sandbagging debates
Four specific decay modes explain why this happens. Catch them early and the system stays alive. Miss them and you're rebuilding from scratch every quarter.
The four decay modes that kill capacity models
Most capacity models don't die from a single catastrophic failure—they decay gradually through four distinct failure patterns. Each has characteristic warning signs and a specific fix protocol.
Model Drift
Capacity assumptions become stale—team changes, allocation shifts, time-off calendar outdated. Scenarios drift from reality and stakeholders lose trust.
Discovery Abandonment
Kanban stops getting updated, team skips discovery for "urgent" work, uncertainty multipliers frozen at initial values. Back to wishful thinking estimates.
Scenario Obsolescence
Last scenario deck presented 3+ months ago, execs make decisions on stale data, nobody can answer "what's at risk if capacity drops 10%?"
Ritual Abandonment
Team stops running the three core rituals—capacity pulse skipped, discovery review forgotten, scenario refresh postponed. Without rituals, the system dies silently.
The next four sections detail each decay mode: what it looks like, warning signs to watch for, and the lightweight ritual that catches it early. If you're already experiencing decay, jump to the rescue protocol at the end.
Decay Mode 1: Model drift
Model drift happens when your capacity assumptions become stale but nobody updates the inputs. Effective capacity still shows 80 eng-weeks/quarter, but Sarah left last month and the backfill starts in Q2. Customer allocation still at 70%, but support tickets doubled after the product launch. Time-off calendar hasn't been updated since January.
Warning Signs You Have Model Drift
- ⚠️ Actual delivery consistently misses scenario predictions by >20%
- ⚠️ Team mentions "but we don't have that many people anymore"
- ⚠️ Customer work allocation feels wrong ("we're spending way more time on support")
- ⚠️ Time-off calendar shows zero PTO for next 2 months (stale data)
The Fix: Weekly Capacity Pulse (5 minutes)
Run a 5-minute capacity pulse check every Friday during standup. Four questions keep the model current without adding ceremony overhead:
- Headcount check: Any team changes this week? (joins, exits, role changes, promotions that shift allocation)
- Allocation drift: Does 70% customer / 30% technical still feel right, or has the mix shifted?
- Calendar sync: Upcoming PTO, conferences, on-call rotations logged in the capacity model?
- Spot-check reality: Does effective capacity match what we're actually delivering this sprint?
Who owns it: Engineering Manager or Delivery Lead (primary), Product Ops (backup). Add to existing Friday standup—last 5 minutes. Update the capacity model in ScopeCone when changes detected.
Pro Tip: Don't wait for "big changes" to update the model. Small drifts compound. Treat capacity like a living document, not a quarterly artifact. A 2-person team change (−10% capacity) might not feel urgent, but if you don't update scenarios immediately, stakeholders will hold you to stale commitments.
Decay Mode 2: Discovery abandonment
Discovery abandonment is the #1 predictor of roadmap blowups. It looks like this: discovery kanban board hasn't moved in 3 weeks, engineers still estimate directly in Jira without discovery stage gates, uncertainty multipliers frozen at initial values (4× → 4× → 4×), product just writes specs and throws them at engineering.
The impact is immediate: estimates revert to wishful thinking (no discovery rigor), scenarios built on bad confidence data, surprise scope explosions mid-sprint because nobody validated assumptions before commitment [2].
Warning Signs You've Abandoned Discovery
- ⚠️ Kanban cards stuck in "Research" for 2+ weeks with no updates
- ⚠️ Team skips discovery for "urgent" work (emergency becomes default mode)
- ⚠️ Engineers surprised by scope during sprint planning ("wait, I thought this was simple")
- ⚠️ Uncertainty multipliers haven't changed in a month (frozen at 4×)
The Fix: Discovery Health Check (10 minutes weekly)
Run a 10-minute discovery health check every Monday. Four questions catch abandonment early:
- Flow audit: Are cards moving through stages (Research → Prototype → Validate → Ready), or stuck?
- Stage gate check: Did we skip gates this week? If yes, why? Was it justified or a regression to old habits?
- Uncertainty trends: Are multipliers dropping as we learn (4× → 2× → 1.25× → 1.1×), or frozen?
- Blockers: What's preventing discovery progress? Missing stakeholder input? Prototype infrastructure not ready?
Who owns it: Product Lead (primary), Design + Tech Lead for bottleneck identification (collaborators). Weekly discovery review, Monday or Friday. If you've abandoned discovery for 3+ weeks, run the recovery protocol:
- Audit current roadmap: Which initiatives have stale estimates?
- Re-run discovery: Move those initiatives back to Research/Prototype stage
- Update scenarios: Flag high-uncertainty work in exec view
- Recommit: Adjust delivery timelines based on real confidence data
Pro Tip: "We're too busy to do discovery" translates to "we're creating uncertainty debt that'll bite us in delivery." Discovery abandonment is the #1 predictor of roadmap blowups. If you only maintain one ritual, make it this one.
Decay Mode 3: Scenario obsolescence
Scenario obsolescence happens when the last scenario deck you presented to execs was Q3 kickoff (3 months ago). Committed/target/stretch bands never updated despite team changes. Stakeholders ask "are we still on track?" and nobody knows. Finance expects the stretch scenario, engineering expects committed, product is confused.
The impact compounds over time: execs make decisions on stale data, you miss opportunities to re-negotiate scope when capacity drops, you over-commit because scenarios don't reflect current reality, and you lose stakeholder trust ("why do we even do this planning exercise?") [1].
Warning Signs Your Scenarios Are Obsolete
- ⚠️ Last scenario update >6 weeks ago
- ⚠️ Team composition or priorities shifted but scenarios didn't
- ⚠️ Stakeholders reference old commitments ("but you said Q2 in the last deck")
- ⚠️ Nobody can answer "what's at risk if we lose 2 engineers next month?"
The Fix: Monthly Scenario Refresh + Trigger Events
Don't wait for quarterly reviews to update scenarios. Refresh when reality changes:
- Capacity shifts >10%: Team changes, allocation rebalance, unexpected leave
- Priority changes: Exec adds "must-have" work mid-quarter
- Discovery reveals scope: Major initiative 2× larger than estimated
- Dependency blocks delivery: External team delay, vendor issue, infra blocker
- Monthly checkpoint: Default refresh even if nothing changed (reality check)
Refresh protocol (20 minutes): Pull latest capacity (headcount, allocation, time-off) → Pull latest discovery (uncertainty multipliers, confidence levels) → Regenerate scenarios (committed/target/stretch) → Delta analysis (what changed vs. last deck?) → Stakeholder brief (if delta >20%, schedule 15-min exec sync)
Communication Template
Scenario Update: Feb 2025
What changed: Sarah's departure moved start date to March (net −10% capacity)
Impact:
- Committed: Feature X delivery slips Q1 → Q2
- Target: Launch on time, but cut Feature Y
- Stretch: Requires backfill hire by March 1
Recommendation: Proceed with Target scenario, de-scope Feature Y
Updated timeline: [link to scenarios in ScopeCone]
Who owns it: Product Lead or EM (whoever presents to execs, primary), Finance (for capacity trade-off conversations, collaborator). Cadence: Monthly + trigger events.
Pro Tip: Don't wait for exec staff meetings to refresh scenarios. Update when reality changes, then communicate proactively. Stakeholders prefer early warnings over surprises. ScopeCone's scenario planner auto-updates when capacity model changes—export the delta and brief leadership within 48 hours.
Decay Mode 4: Ritual abandonment
Ritual abandonment is the silent killer. It looks like this: capacity pulse gets skipped because "we're too busy shipping," discovery health check falls off the calendar, scenario refresh gets postponed indefinitely. The capacity model doesn't break loudly—it just stops being used. Three months later, nobody can remember when they last updated it.
The impact compounds quickly: capacity shifts but scenarios don't update, discovery stalls but nobody notices, predictability drops but the team blames "bad estimates" instead of fixing the system. Research shows teams with structured capacity management see 60% better delivery predictability [1]—but only if the rituals actually run.
Warning Signs You've Abandoned Rituals
- ⚠️ Can't remember when you last ran capacity pulse
- ⚠️ Discovery board timestamps show 3+ weeks since last update
- ⚠️ Scenario deck presented to execs is 2+ months old
- ⚠️ Team says "we should really update the capacity model" but never does
The Fix: Process Health Metrics from Your Rituals
Don't track vanity metrics on a dashboard nobody checks. Instead, track three process health signals that are byproducts of rituals you're already running. These aren't separate tracking—they emerge naturally from your weekly capacity pulse, discovery health check, and retrospectives.
The 3 Process Health Metrics
Capacity Delta
What it measures: How much effective capacity changed this week (from Friday Capacity Pulse)
Threshold: >10% change week-over-week
Action: Trigger scenario refresh immediately—stakeholders are making decisions on stale data
Discovery Throughput
What it measures: Cards moving through discovery stages (from Monday Discovery Health Check)
Threshold: <2 cards/week for 2+ weeks
Action: Discovery stalled—remove blockers or deprioritize stuck work
Delivery Predictability
What it measures: Did we deliver what we committed in our scenarios?
Threshold: Missing committed scenario by >20%
Action: Run 5 Whys retrospective—was it capacity model? discovery? external blocker? Fix the root cause
| Metric | Source | Threshold | Action |
|---|---|---|---|
| Capacity Delta | Friday Capacity Pulse | >10% weekly change | Refresh scenarios immediately |
| Discovery Throughput | Monday Discovery Check | <2 cards/week for 2+ weeks | Remove blockers, deprioritize stuck work |
| Delivery Predictability | Sprint/quarter retrospective | Miss committed by >20% | 5 Whys to find root cause, fix system |
Why these three? They're leading and lagging indicators of system health. Capacity Delta and Discovery Throughput catch problems early (rituals running but producing warning signals). Delivery Predictability catches systemic failures (you're hitting scenarios consistently or something's broken—run 5 Whys to find the root cause).
Weekly Process Health Check (5 minutes): Review the three metrics during Monday planning. Any red thresholds? Execute the action. Trends worsening? Investigate root cause. This isn't separate metric tracking—it's reviewing the outputs your rituals already produce.
Who owns it: EM or Product Lead (primary), whoever runs Monday planning. Cadence: Weekly review, built into existing planning meeting.
Pro Tip: These metrics tell you why rituals get abandoned. Capacity Delta shows when your model becomes irrelevant. Discovery Throughput shows when your team stops trusting the process. Delivery Predictability shows when the whole system needs fixing. Track the process, not the people [3].
The minimal viable maintenance cadence
You don't need more meetings. The goal isn't ceremony—it's integration. Bolt these checks into existing rituals:
Friday: Capacity Pulse (5 min)
Check headcount changes, allocation drift (does 70/30 split still feel right?), upcoming PTO/conferences, and spot-check if effective capacity matches actual delivery.
Monday: Discovery Health Check (5 min)
Audit card movement through kanban stages, verify stage gates weren't skipped, check if uncertainty multipliers are dropping as expected, identify blockers.
Monday: Process Health Check (5 min)
Review the 3 process metrics: Capacity Delta from Friday pulse, Discovery Throughput from this check, Delivery Predictability from last retro. Any red thresholds? Execute the action protocol immediately.
Full Maintenance Calendar
- Friday standup +5 min: Capacity pulse (headcount, allocation, calendar)
- Monday planning +5 min: Discovery health check (cards moving? gates honored?)
- Monday planning +5 min: Process health check (review 3 metrics, any reds?)
- Scenario refresh: Pull latest capacity + discovery, regenerate bands, check for >20% delta
- If delta significant → 15-min stakeholder brief
- Model recalibration: Full audit of capacity assumptions, allocation accuracy, metric thresholds
- Retrospective: What decay did we catch? What did we miss? Adjust rituals accordingly
Total time investment: Weekly: 15 min (0.6% of a 40-hour week) | Monthly: +20 min | Quarterly: +60 min
Compare to cost of decay: Roadmap blowup: 8-20 hours of re-planning, stakeholder damage control | Missed commitments: Lost exec trust, budget cuts, team morale hit | Stale model rescue: 10+ hours rebuilding from scratch
The Trade-Off: 15 min/week prevents decay. 10+ hours rescues a dead system. Prevention wins every time.
Turn this maintenance cost into capacity
Map the dollars you just calculated into real planning slots. Build a shared capacity model, compare scenarios, and decide what debt to attack without guessing.
Making it stick: ownership & integration
Maintenance fails when there's no clear owner. Common anti-patterns: "The team owns it" (translation: nobody owns it), "We'll figure it out" (translation: we won't), "Product's job" vs "Engineering's job" (translation: both assume the other is doing it).
Ownership Model
Capacity Pulse
Primary: EM or Delivery Lead | Backup: Product Ops | Cadence: Weekly (Friday)
Discovery Health Check
Primary: Product Lead | Backup: Design Lead | Cadence: Weekly (Monday)
Process Health Check
Primary: EM or Product Lead | Backup: Tech Lead | Cadence: Weekly (Monday)
Scenario Refresh
Primary: Product Lead or EM | Backup: Finance Partner | Cadence: Monthly + triggers
Model Recalibration
Primary: EM + Product Lead | Backup: — | Cadence: Quarterly
Integration Checklist
- ✓Add to existing meetings — Don't create new ceremonies, bolt onto standup/planning
- ✓Assign explicit owners — Name in RACI, not "the team"
- ✓Document the protocol — 1-page ritual guide, share with team
- ✓Set calendar reminders — Capacity pulse (Fridays), discovery health check (Mondays), scenarios (1st of month)
- ✓Link to tools — ScopeCone capacity model, discovery board, scenario planner
- ✓Review in retros — "Did we maintain the model this sprint?" Yes/No/Why not?
The First 30 Days: Week 1: Assign owners, add rituals to calendar. Week 2-4: Run all three weekly checks, gather feedback. End of Month 1: First scenario refresh, retrospective on what's working.
After 3 Months: If the rituals feel like extra work, something's wrong. They should feel like "this is how we work" not "this is work on top of work."
Rescue protocol: your model is already dead
If you're reading this and thinking "yep, that's us"—you're not alone. Most capacity models die within 8 weeks. Here's the rescue path:
Week 1: Audit Current State
- □ When did you last update capacity model? (headcount, allocation, time-off)
- □ When did you last move discovery cards? (check timestamps)
- □ When did you last present scenarios to execs? (find the deck)
- □ When did you last run all three rituals? (honestly?)
Week 2: Rebuild Trust
Week 3-4: Install Maintenance Rituals
- □ Assign owners for capacity pulse, discovery check, process health check
- □ Add to existing meeting calendar (don't create net-new meetings)
- □ Run first monthly scenario refresh
- □ Document the protocol (1-page guide)
Month 2+: Prevent Re-Decay — Use the warning signs cheat sheet monthly. Retrospective: "Did we maintain the model?" Adjust rituals based on what's working.
The Hard Truth: If you rescue a dead model but don't install maintenance rituals, it'll die again in 6 weeks. Prevention must become habit, not a one-time fix.
Three habits that prevent decay
The minimal viable maintenance system:
- Weekly Capacity Pulse (5 min)
Check headcount, allocation, time-off. Update model when changes detected. Owner: EM or Delivery Lead. - Weekly Discovery Health Check (5 min)
Audit card movement, stage gates, uncertainty trends. Enforce discovery rigor, catch abandonment early. Owner: Product Lead. - Weekly Process Health Check (5 min)
Review 3 metrics from rituals: Capacity Delta, Discovery Throughput, Delivery Predictability. Any reds? Execute action. Owner: EM or Product Lead. - Monthly Scenario Refresh (20 min)
Pull latest capacity + discovery data, regenerate committed/target/stretch bands, brief stakeholders if delta >20%. Owner: Product Lead or EM.
Total Time: 15 min/week + 20 min/month
The Return: Prevent roadmap blowups (8-20 hours saved), maintain stakeholder trust (priceless), keep team aligned on reality vs. wishful thinking.
Your Next Action
- Audit your current state — Which decay modes do you have? (Use warning signs in sections above)
- Assign owners — Who runs capacity pulse? Discovery check? Scenario refresh?
- Block calendar time — Add rituals to existing meetings this week
- Run first check — Start with the capacity pulse (5 min, right now)