diff --git a/docs/analytics.md b/docs/analytics.md
index c99d8cf..e78b61f 100644
--- a/docs/analytics.md
+++ b/docs/analytics.md
@@ -10,10 +10,17 @@ scripting.
 ```bash
 ytstudio analytics overview                          # channel summary for the last 28 days
 ytstudio analytics overview --days 7                 # rolling 7-day window
+ytstudio analytics overview --no-compare             # totals only, no growth deltas
 ytstudio analytics video <video-id>                  # one video's recent performance
 ytstudio analytics video <video-id> --days 90        # longer window for one video
 ```
 
+By default `overview` also compares each metric to the previous equal-length
+window and shows the percent change (green up, red down). Pass `--no-compare`
+for plain totals. In JSON output the comparison is exposed as `previous`
+(the prior window's metrics) and `pct_change` (percent change per metric,
+`null` when there is no prior-window baseline).
+
 ## Custom queries
 
 `analytics query` is a thin wrapper over the YouTube Analytics
@@ -55,7 +62,7 @@ Run `ytstudio analytics --help` (or open the
 === "JSON"
 
     ```bash
-    ytstudio analytics overview -o json | jq '.rows[0]'
+    ytstudio analytics overview -o json | jq '.pct_change'
     ```
 
 ## Quota
diff --git a/src/ytstudio/commands/analytics.py b/src/ytstudio/commands/analytics.py
index 2fad1fd..6db7962 100644
--- a/src/ytstudio/commands/analytics.py
+++ b/src/ytstudio/commands/analytics.py
@@ -79,9 +79,44 @@ def fetch_query(
     return api(analytics_service.reports().query(**query_params))
 
 
+# Metrics shown with a period-over-period delta in the overview.
+OVERVIEW_COMPARE_METRICS = (
+    "views",
+    "estimatedMinutesWatched",
+    "averageViewDuration",
+    "likes",
+    "comments",
+)
+
+
+def _pct_change(current: float, previous: float) -> float | None:
+    """Percent change vs the previous window, or None when there is no baseline."""
+    if previous == 0:
+        return None
+    return round((current - previous) / previous * 100, 1)
+
+
+def _delta_suffix(metrics: dict, previous: dict | None, metric: str) -> str:
+    """Coloured ' (+N%)' suffix comparing a metric to the previous window."""
+    if previous is None:
+        return ""
+    current = float(metrics.get(metric, 0))
+    pct = _pct_change(current, float(previous.get(metric, 0)))
+    if pct is None:
+        return dim(" (new)") if current else ""
+    colour = "green" if pct >= 0 else "red"
+    sign = "+" if pct >= 0 else ""
+    return f" [{colour}]({sign}{pct:.0f}%)[/{colour}]"
+
+
 @app.command()
 def overview(
     days: int = typer.Option(28, "--days", "-d", help="Number of days to analyze"),
+    compare: bool = typer.Option(
+        True,
+        "--compare/--no-compare",
+        help="Compare each metric to the previous equal-length window",
+    ),
     output: str = typer.Option("table", "--output", "-o", help="Output format: table, json"),
 ):
     """Get channel overview analytics"""
@@ -119,8 +154,43 @@ def overview(
 
     metrics = dict(zip(headers, rows[0], strict=False))
 
+    # Previous equal-length window ending the day before the current one starts,
+    # so the two windows do not share a boundary day.
+    previous = None
+    pct_change = None
+    if compare:
+        prev_end = (datetime.now() - timedelta(days=days + 1)).strftime("%Y-%m-%d")
+        prev_start = (datetime.now() - timedelta(days=days * 2 + 1)).strftime("%Y-%m-%d")
+        prev_response = fetch_query(
+            data_service,
+            analytics_service,
+            metric_names=metric_names,
+            dimension_names=[],
+            start_date=prev_start,
+            end_date=prev_end,
+            days=days,
+        )
+        prev_rows = prev_response.get("rows", [])
+        if prev_rows:
+            prev_headers = [h["name"] for h in prev_response.get("columnHeaders", [])]
+            previous = dict(zip(prev_headers, prev_rows[0], strict=False))
+            pct_change = {
+                m: _pct_change(float(metrics.get(m, 0)), float(previous.get(m, 0)))
+                for m in OVERVIEW_COMPARE_METRICS
+            }
+
     if output == "json":
-        print(json.dumps({"analytics": metrics, "days": days}, indent=2))
+        print(
+            json.dumps(
+                {
+                    "analytics": metrics,
+                    "days": days,
+                    "previous": previous,
+                    "pct_change": pct_change,
+                },
+                indent=2,
+            )
+        )
         return
 
     views = int(metrics.get("views", 0))
@@ -131,16 +201,23 @@ def overview(
     likes = int(metrics.get("likes", 0))
     comments = int(metrics.get("comments", 0))
 
-    console.print(f"\n[bold]Channel Analytics[/bold] {dim(f'(last {days} days)')}\n")
+    subtitle = f"(last {days} days"
+    subtitle += f" vs previous {days})" if previous is not None else ")"
+    console.print(f"\n[bold]Channel Analytics[/bold] {dim(subtitle)}\n")
     table = create_kv_table()
 
-    table.add_row(dim("views"), format_number(views))
-    table.add_row(dim("watch time"), f"{watch_hours} hours")
-    table.add_row(dim("avg duration"), f"{avg_secs // 60}:{avg_secs % 60:02d}")
+    def d(metric: str) -> str:
+        return _delta_suffix(metrics, previous, metric)
+
+    table.add_row(dim("views"), format_number(views) + d("views"))
+    table.add_row(dim("watch time"), f"{watch_hours} hours" + d("estimatedMinutesWatched"))
+    table.add_row(
+        dim("avg duration"), f"{avg_secs // 60}:{avg_secs % 60:02d}" + d("averageViewDuration")
+    )
     table.add_row(dim("subscribers gained"), f"[green]+{subs_gained}[/green]")
     table.add_row(dim("subscribers lost"), f"[red]-{subs_lost}[/red]")
-    table.add_row(dim("likes"), format_number(likes))
-    table.add_row(dim("comments"), format_number(comments))
+    table.add_row(dim("likes"), format_number(likes) + d("likes"))
+    table.add_row(dim("comments"), format_number(comments) + d("comments"))
 
     console.print(table)
 
diff --git a/tests/test_analytics.py b/tests/test_analytics.py
index 1d1c2c6..d84fbfd 100644
--- a/tests/test_analytics.py
+++ b/tests/test_analytics.py
@@ -97,6 +97,112 @@ def test_overview_no_data(self):
             assert result.exit_code == 0
             assert "No analytics data" in result.output
 
+    @staticmethod
+    def _overview_response(row):
+        return {
+            "columnHeaders": [
+                {"name": "views"},
+                {"name": "estimatedMinutesWatched"},
+                {"name": "averageViewDuration"},
+                {"name": "subscribersGained"},
+                {"name": "subscribersLost"},
+                {"name": "likes"},
+                {"name": "comments"},
+            ],
+            "rows": [row],
+        }
+
+    def _two_window_services(self, current_row, previous_row):
+        data_service = MagicMock()
+        analytics_service = MagicMock()
+        data_service.channels.return_value.list.return_value.execute.return_value = {
+            "items": [{"id": "UC_test"}]
+        }
+        analytics_service.reports.return_value.query.return_value.execute.side_effect = [
+            self._overview_response(current_row),
+            self._overview_response(previous_row),
+        ]
+        return data_service, analytics_service
+
+    def test_overview_compare_deltas_table(self):
+        # views 12000 vs 10000 -> +20%, avg duration 180 vs 200 -> -10%
+        data_svc, analytics_svc = self._two_window_services(
+            [12000, 6000, 180, 42, 3, 770, 25],
+            [10000, 6000, 200, 42, 3, 700, 25],
+        )
+        with (
+            patch("ytstudio.commands.analytics.get_data_service", return_value=data_svc),
+            patch("ytstudio.commands.analytics.get_analytics_service", return_value=analytics_svc),
+        ):
+            result = runner.invoke(app, ["analytics", "overview"])
+            assert result.exit_code == 0
+            assert "+20%" in result.output
+            assert "-10%" in result.output
+
+    def test_overview_compare_json(self):
+        data_svc, analytics_svc = self._two_window_services(
+            [12000, 6000, 180, 42, 3, 770, 25],
+            [10000, 6000, 200, 42, 3, 700, 25],
+        )
+        with (
+            patch("ytstudio.commands.analytics.get_data_service", return_value=data_svc),
+            patch("ytstudio.commands.analytics.get_analytics_service", return_value=analytics_svc),
+        ):
+            result = runner.invoke(app, ["analytics", "overview", "-o", "json"])
+            assert result.exit_code == 0
+            payload = json.loads(result.output)
+            assert payload["previous"]["views"] == 10000
+            assert payload["pct_change"]["views"] == 20.0
+            assert payload["pct_change"]["averageViewDuration"] == -10.0
+            assert payload["pct_change"]["likes"] == 10.0
+
+    def test_overview_compare_windows_do_not_overlap(self):
+        # the previous window must end before the current window starts
+        data_svc, analytics_svc = self._two_window_services(
+            [12000, 6000, 180, 42, 3, 770, 25],
+            [10000, 6000, 200, 42, 3, 700, 25],
+        )
+        with (
+            patch("ytstudio.commands.analytics.get_data_service", return_value=data_svc),
+            patch("ytstudio.commands.analytics.get_analytics_service", return_value=analytics_svc),
+        ):
+            result = runner.invoke(app, ["analytics", "overview", "--days", "7"])
+            assert result.exit_code == 0
+
+        calls = analytics_svc.reports.return_value.query.call_args_list
+        current_start = calls[0].kwargs["startDate"]
+        previous_end = calls[1].kwargs["endDate"]
+        assert previous_end < current_start
+
+    def test_overview_no_compare(self):
+        data_svc, analytics_svc = self._mock_overview_services()
+        with (
+            patch("ytstudio.commands.analytics.get_data_service", return_value=data_svc),
+            patch("ytstudio.commands.analytics.get_analytics_service", return_value=analytics_svc),
+        ):
+            result = runner.invoke(app, ["analytics", "overview", "--no-compare", "-o", "json"])
+            assert result.exit_code == 0
+            payload = json.loads(result.output)
+            assert payload["previous"] is None
+            assert payload["pct_change"] is None
+            # only the current window is queried
+            assert analytics_svc.reports.return_value.query.return_value.execute.call_count == 1
+
+    def test_overview_compare_zero_previous(self):
+        # previous window has no data -> guard against divide-by-zero
+        data_svc, analytics_svc = self._two_window_services(
+            [500, 6000, 180, 42, 3, 789, 25],
+            [0, 0, 0, 0, 0, 0, 0],
+        )
+        with (
+            patch("ytstudio.commands.analytics.get_data_service", return_value=data_svc),
+            patch("ytstudio.commands.analytics.get_analytics_service", return_value=analytics_svc),
+        ):
+            result = runner.invoke(app, ["analytics", "overview", "-o", "json"])
+            assert result.exit_code == 0
+            payload = json.loads(result.output)
+            assert payload["pct_change"]["views"] is None
+
     def test_video_not_found(self, mock_auth):
         mock_auth.videos.return_value.list.return_value.execute.return_value = {"items": []}
         with (