diff --git a/docs/docs/in_depth/feature-group-matching.md b/docs/docs/in_depth/feature-group-matching.md
index 0929e6f1..5afeb462 100644
--- a/docs/docs/in_depth/feature-group-matching.md
+++ b/docs/docs/in_depth/feature-group-matching.md
@@ -25,6 +25,8 @@ def match_feature_group_criteria(cls, feature_name, options, data_access_collect
 
 Do not call `FeatureChainParser.match_configuration_feature_chain_parser` directly from a match hook: it raises on an option value the `PROPERTY_MAPPING` rejects, and an exception out of the hook aborts feature identification for every candidate, not just yours. `match_parser_criteria` turns that rejection into a non-match, and the reason still reaches the user in the "No feature groups found" error.
 
+The options view depends on the caller: feature resolution passes declared (pre-default) options, while filter matching runs after intake and passes the resolved feature's effective (post-default) options merged onto the filter feature's own. Matching logic that reads option values can see different values on the two paths. See [Applying declared defaults](property-mapping.md#applying-declared-defaults).
+
 ### 2. PROPERTY_MAPPING Configuration
 
 The `PROPERTY_MAPPING` defines how configuration-based features are validated:
diff --git a/docs/docs/in_depth/property-mapping.md b/docs/docs/in_depth/property-mapping.md
index 33d35134..88ef5123 100644
--- a/docs/docs/in_depth/property-mapping.md
+++ b/docs/docs/in_depth/property-mapping.md
@@ -378,7 +378,8 @@ For an engine-driven request intake has already run, so the compute-boundary cal
 no-op. It carries the work only for direct `FeatureSet` use that bypasses the engine, and there the
 collapsing twins raise instead of merging.
 
-Both sites run *after* resolution, so materialization never changes **matching**. It does change
+Both sites run *after* resolution, so materialization never changes **resolution matching** (filter
+matching runs after intake and does observe materialized values, see below). It does change
 how features **group**: intake materialization deliberately canonicalizes default-equivalent twins,
 so a feature that passes a declared default explicitly and one that omits it become equal and merge
 into a single feature (with a warning naming the duplicated request) instead of computing twice.
@@ -393,7 +394,7 @@ Which stage sees which view of the options:
 
 | Stage | Options view |
 | --- | --- |
-| Parse, bind, match, resolve (subtype resolution applies defaults internally) | declared (pre-default) |
+| Parse, bind, resolution match, resolve (subtype resolution applies defaults internally) | declared (pre-default) |
 | `input_features()` and child option inheritance | declared (pre-default) |
 | Splitting, planning, filter matching, `validate_input_features`, compute | effective (post-default) |
 
@@ -402,8 +403,9 @@ path the safety net materializes inside `run_calculate_feature`, which is after
 `validate_input_features`, not before it.
 
 Filter matching is the one place the same classmethod sees both views: `GlobalFilter` calls
-`match_feature_group_criteria` again after intake, so it observes effective options where feature
-resolution saw declared ones.
+`match_feature_group_criteria` again after intake, passing the filter feature's own options
+enriched from the resolved feature's effective ones (`unify_options`), so a key the filter feature
+omits arrives materialized where feature resolution saw it declared.
 
 ``` python
 graph_type = feature.options.get("graph_type")  # the declared default when the caller omitted it
diff --git a/mloda/core/abstract_plugins/components/feature_chainer/feature_chain_parser_mixin.py b/mloda/core/abstract_plugins/components/feature_chainer/feature_chain_parser_mixin.py
index fd656bc9..71b4f370 100644
--- a/mloda/core/abstract_plugins/components/feature_chainer/feature_chain_parser_mixin.py
+++ b/mloda/core/abstract_plugins/components/feature_chainer/feature_chain_parser_mixin.py
@@ -247,6 +247,9 @@ def match_feature_group_criteria(
         variable ``MLODA_ALLOW_FORWARDED_NAME_MISMATCH=1`` downgrades this error to a
         warning.
 
+        The options view depends on the caller: feature resolution passes declared (pre-default)
+        options, filter matching a post-intake merge. See ``FeatureGroup.match_feature_group_criteria``.
+
         Args:
             feature_name: Feature name to match
             options: Options containing configuration
diff --git a/mloda/core/abstract_plugins/feature_group.py b/mloda/core/abstract_plugins/feature_group.py
index 6e6a218c..8946550d 100644
--- a/mloda/core/abstract_plugins/feature_group.py
+++ b/mloda/core/abstract_plugins/feature_group.py
@@ -602,6 +602,13 @@ def match_feature_group_criteria(
         You can disallow them by removing them. However, often you can just use the default.
         If you want to implement a concrete implementation, e.g. just accept specific names,
         then you can overwrite this function.
+
+        The options view depends on the caller. Feature resolution passes declared (pre-default)
+        options. GlobalFilter matches a filter feature after intake, passing that feature's own
+        options enriched from the resolved feature's effective (post-default) ones: a key the filter
+        feature declares itself keeps its declared value, an omitted key arrives materialized.
+        Matching logic that reads option values can therefore see different values on the two paths.
+        See ``docs/in_depth/property-mapping.md`` ("Applying declared defaults").
         """
 
         base_feature_name = cls.get_column_base_feature(feature_name)