Improve tasks and backdate insights

core/views/compose.py

@@ -1228,7 +1228,7 @@ def _trend_meta(current, previous, higher_is_better=True):
     }
 
 
-def _emotion_meta(metric_kind, value):
+def _emotion_meta(metric_kind, value, metric_key=None):
     score = _to_float(value)
     if score is None:
         return {
@@ -1239,6 +1239,25 @@ def _emotion_meta(metric_kind, value):
     if metric_kind == "confidence":
         score = score * 100.0
     if metric_kind == "count":
+        key = str(metric_key or "").strip().lower()
+        if key == "sample_days":
+            if score >= 14:
+                return {
+                    "icon": "fa-solid fa-calendar-check",
+                    "class_name": "has-text-success",
+                    "label": "Broad Coverage",
+                }
+            if score >= 5:
+                return {
+                    "icon": "fa-solid fa-calendar-days",
+                    "class_name": "has-text-warning",
+                    "label": "Adequate Coverage",
+                }
+            return {
+                "icon": "fa-solid fa-calendar-xmark",
+                "class_name": "has-text-danger",
+                "label": "Narrow Coverage",
+            }
         if score >= 80:
             return {
                 "icon": "fa-solid fa-chart-column",
@@ -1433,7 +1452,7 @@ def _quick_insights_rows(conversation):
         point_count = conversation.metric_snapshots.exclude(
             **{f"{field_name}__isnull": True}
         ).count()
-        emotion = _emotion_meta(spec["kind"], current)
+        emotion = _emotion_meta(spec["kind"], current, spec["key"])
         rows.append(
             {
                 "key": spec["key"],
@@ -4035,6 +4054,7 @@ class ComposeQuickInsights(LoginRequiredMixin, View):
                     "Arrow color indicates improving or risk direction for that metric.",
                     "State uses participant feedback (Withdrawing/Overextending/Balanced) when available.",
                     "Values are computed from all linked platform messages for this person.",
+                    "Data labels are metric-specific (for example, day coverage is rated separately from message volume).",
                     "Face indicator maps value range to positive, mixed, or strained climate.",
                     "Use this card for fast triage; open AI Workspace for full graphs and details.",
                 ],