diff --git a/docs/product/SCENARIO_EPIC_MIGRATION.md b/docs/product/SCENARIO_EPIC_MIGRATION.md
new file mode 100644
index 0000000..058787c
--- /dev/null
+++ b/docs/product/SCENARIO_EPIC_MIGRATION.md
@@ -0,0 +1,206 @@
+# Scenario Epic Migration
+
+Last updated: 2026-05-19
+
+## Summary
+
+This migration establishes scenario-based product epics without reorganizing the whole backlog. It follows the accepted proposal in `docs/product/END_TO_END_EPICS_PROPOSAL.md` and intentionally keeps existing technical issues, labels, and discussions intact.
+
+The migration is conservative:
+
+- created only the six agreed top-level scenario epics
+- preserved existing issue history
+- did not close, delete, mass rename, or relabel issues
+- did not modify completed issues
+- kept deterministic product loops separate from research and future ML work
+
+## Created epics
+
+| Scenario epic | Issue |
+| --- | --- |
+| `[SCENARIO] Morning Readiness Loop` | #91 |
+| `[SCENARIO] Explainable Readiness Experience` | #92 |
+| `[SCENARIO] Ride Preparation and Recommendation Loop` | #93 |
+| `[SCENARIO] Post-Workout Feedback Loop` | #94 |
+| `[SCENARIO] Readiness Calibration Loop` | #95 |
+| `[SCENARIO] Research Sandbox` | #96 |
+
+Each scenario epic contains:
+
+- problem statement
+- user-facing flow
+- definition of done
+- linked existing issues
+- migration note explaining that the change is additive
+
+All six were created with the existing `type:epic` label only. No new labels were introduced.
+
+## Issue mappings
+
+### Morning Readiness Loop (#91)
+
+Linked issues:
+
+- #7 iOS app as HealthKit ingestion layer
+- #51 iOS onboarding flow
+- #52 System status & sync screen
+- #88 Event-driven morning briefing after HealthKit sync
+
+Migration action:
+
+- created new scenario epic #91
+- did not rename or close any linked issue
+
+### Explainable Readiness Experience (#92)
+
+Linked issues:
+
+- #37 Readiness stability and volatility analytics
+- #50 Readiness explanation layer
+- #53 Readiness, fatigue, and recovery trend visualization
+- #56 Metrics layer & explainability
+- #57 Extract metrics layer
+- #58 Metric-level documentation
+- #59 iOS explainable readiness presentation
+- #60 Unit tests for core metrics
+- #81 Dynamic semantic slogan system
+
+Migration action:
+
+- created new scenario epic #92
+- added a migration comment to #56 pointing future planning to #92
+- did not rename #56, even though it is technical, to avoid unnecessary churn
+
+### Ride Preparation and Recommendation Loop (#93)
+
+Linked issues:
+
+- #1 Athlete profile
+- #13 Bike profile
+- #14 Equipment tracking
+- #15 Service log
+- #16 Athlete Profile epic
+- #17 Equipment & maintenance system epic
+- #18 Bike Profile epic
+- #19 Maintenance alerts
+- #20 Equipment status visualization
+- #22 Integrate equipment status into ride briefing
+- #55 Training outcome prediction layer
+- #76 Recommendation engine v1
+
+Migration action:
+
+- created new scenario epic #93
+- added migration comments to #16, #17, and #18 pointing future planning to #93
+- did not close or merge the older profile/equipment epics
+- did not rename #55 or #76 yet, even though clearer names were suggested in the proposal
+
+### Post-Workout Feedback Loop (#94)
+
+Linked issues:
+
+- #74 Morning recovery feedback collection
+- #75 Pre-ride subjective readiness feedback
+- #82 Unified Subjective Feedback Model epic
+- #85 Sport-aware optional second-tap feedback
+- #86 Source-aware subjective feedback ingestion
+- #87 Subjective feedback ontology and versioning
+- #89 Scheduled next-day recovery feedback prompt
+
+Completed prior work referenced but not modified:
+
+- #83 Extensible subjective feedback schema
+- #84 Next-day recovery subjective feedback
+
+Migration action:
+
+- created new scenario epic #94
+- added a migration comment to #82 pointing future planning to #94
+- did not modify completed issues #83 or #84
+- did not close #74 as duplicate, because that should be a manual follow-up after confirming overlap with #84 and #89
+
+### Readiness Calibration Loop (#95)
+
+Linked issues:
+
+- #54 Readiness calibration with subjective outcomes
+- #73 Adaptive Feedback Loop epic
+- #77 Prediction vs outcome calibration analytics
+- #78 Persist feature snapshots for future ML datasets
+- #79 Dataset export pipeline for calibration and ML
+- #80 Subjective feedback analytics
+
+Migration action:
+
+- created new scenario epic #95
+- added a migration comment to #73 pointing future planning to #95
+- explicitly separated production calibration from speculative/offline research
+- did not split #78 or rename #79 yet
+
+### Research Sandbox (#96)
+
+Linked issues:
+
+- #36 Automated model experimentation sandbox
+- future research portions of #78 and #79 after deterministic calibration export is stable
+
+Migration action:
+
+- created new scenario epic #96
+- added a migration comment to #36 pointing future research planning to #96
+- kept research offline and separate from deterministic readiness/recommendation behavior
+
+### Ambiguous enabling work
+
+Issue #9 Unified Load Model was not forced into one scenario epic. It spans:
+
+- Morning Readiness Loop, because whole-day load affects the morning answer
+- Ride Preparation and Recommendation Loop, because total load affects recommendation safety
+- Readiness Calibration Loop, because changing load inputs should be validated against outcomes
+
+Migration action:
+
+- added a migration comment to #9 explaining that it remains enabling work for #91, #93, and #95
+- did not rename, close, split, or relabel #9
+
+## Untouched ambiguous areas
+
+The following areas were intentionally left unchanged:
+
+- existing labels and auto-label workflow
+- project views and milestones
+- completed issues, including #83 and #84
+- older technical epics, except for additive comments
+- issue titles that may eventually benefit from clearer naming
+- unified load work (#9-#12), because it crosses multiple scenario loops
+- research/dataset wording in #78 and #79, because splitting those should happen with implementation planning
+- UI identity work (#81), because it may support explainability aesthetics but should not drive the core product loops
+
+## Suggested manual follow-ups
+
+1. Decide whether older technical epics should remain open as umbrella context or be closed as superseded after the scenario epics are accepted.
+2. Review whether #74 is now superseded by completed #84 plus open #89.
+3. Consider renaming #56 to make it clearly subordinate to #92, or leave it as an implementation umbrella.
+4. Consider renaming #55 to rule-based pre-ride outcome classification to reduce ML/prediction ambiguity.
+5. Consider renaming #76 to deterministic recommendation object and reason codes.
+6. Decide whether #78 should be split into production snapshots and future ML dataset enrichment.
+7. Decide whether #79 should be renamed to calibration dataset export pipeline.
+8. Decide whether new scenario epics need additional area labels after the first planning pass; this migration only used `type:epic`.
+
+## Risks avoided intentionally
+
+- No mass issue renames, because title churn makes backlog history harder to follow.
+- No mass closures, because duplicate/superseded status should be confirmed issue by issue.
+- No label deletion or new taxonomy, because current area labels remain useful for filtering.
+- No completed issue edits, because completed feedback work is historical evidence.
+- No project view changes, because views should be updated after the scenario epics are reviewed in GitHub.
+- No speculative ML integration, because calibration and research remain separate from the deterministic product core.
+- No forced mapping for unified load, because it is an enabling model stream that affects multiple user loops.
+
+## Verification
+
+- Confirmed no open scenario epics with the requested `[SCENARIO] ...` titles existed before creation.
+- Created the six requested scenario epics as #91-#96.
+- Added additive migration comments only to open legacy epic/research issues: #9, #16, #17, #18, #36, #56, #73, and #82.
+- Did not modify completed issues.
+- Did not reorganize project views.