diff --git a/MIGRATION_v3_to_v4.md b/MIGRATION_v3_to_v4.md
index 7b406e56..07840d04 100644
--- a/MIGRATION_v3_to_v4.md
+++ b/MIGRATION_v3_to_v4.md
@@ -48,13 +48,16 @@ dependencies = [
 
 ## Fix removed-type imports first — they cascade
 
-Real-adopter feedback (salesagent v3→v4 experiment, 270 files scanned, 161
-test-collection failures): consumers tend to centralize SDK imports in one
-schema module, so a single broken import there crashes test collection across
-the whole codebase. salesagent re-exported through `src/core/schemas/_base.py`
-— **one missing `FormatCategory` import there cascaded into ~140 test failures
-during pytest collect-only**, and stubbing it revealed the next ~140-test
-cascade from `BrandManifest`, then the next from the `generated_poc` reach-ins.
+Real-adopter feedback (salesagent v3→v4 experiment at v4.0 release): 270 files
+scanned, 161 test-collection failures. These figures reflect tooling at initial
+release, before subsequent alias coverage and codemod improvements; run the codemod against
+your own tree to see your actual count. Consumers tend to centralize SDK imports
+in one schema module, so a single broken import there crashes test collection
+across the whole codebase. salesagent re-exported through
+`src/core/schemas/_base.py` — **one missing `FormatCategory` import there
+cascaded into ~140 test failures during pytest collect-only**, and stubbing it
+revealed the next ~140-test cascade from `BrandManifest`, then the next from
+the `generated_poc` reach-ins.
 
 The codemod's static finding count understates this by 100x+. To minimize the
 felt blast radius, work in this order: