diff --git a/CHANGELOG.md b/CHANGELOG.md
index 05c52daa1..9d6406860 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -4,6 +4,7 @@
 * [Feature] Using native DuckDB `.df()` method when using `autopandas` 
 
 * [Doc] documenting `%sqlcmd tables`/`%sqlcmd columns`
+* [Feature] Better error messages when function used in plotting API unsupported by DB driver (#159)
 
 ## 0.7.4 (2023-04-28)
 No changes
diff --git a/src/sql/plot.py b/src/sql/plot.py
index e8c2be9cf..e6c55e842 100644
--- a/src/sql/plot.py
+++ b/src/sql/plot.py
@@ -6,6 +6,8 @@
 from jinja2 import Template
 
 from sql.util import flatten
+from sqlalchemy.exc import ProgrammingError
+from sql import exceptions
 
 try:
     import matplotlib.pyplot as plt
@@ -30,6 +32,7 @@ def _summary_stats(conn, table, column, with_=None):
 
     if not conn:
         conn = sql.connection.Connection.current
+    driver = conn._get_curr_sqlalchemy_connection_info()["driver"]
 
     template = Template(
         """
@@ -44,7 +47,16 @@ def _summary_stats(conn, table, column, with_=None):
 
     query = template.render(table=table, column=column)
 
-    values = conn.execute(query, with_).fetchone()
+    try:
+        values = conn.execute(query, with_).fetchone()
+    except ProgrammingError as e:
+        print(e)
+        raise exceptions.RuntimeError(
+            f"\nEnsure that percentile_disc function is available on {driver}."
+        )
+    except Exception as e:
+        raise e
+
     keys = ["q1", "med", "q3", "mean", "N"]
     return {k: float(v) for k, v in zip(keys, flatten(values))}
 
diff --git a/src/tests/integration/test_mssql.py b/src/tests/integration/test_mssql.py
index 63057ea14..4bd8a3f72 100644
--- a/src/tests/integration/test_mssql.py
+++ b/src/tests/integration/test_mssql.py
@@ -1,6 +1,7 @@
 import pyodbc
 import pytest
 from matplotlib import pyplot as plt
+from IPython.core.error import UsageError
 
 
 def test_query_count(ip_with_MSSQL, test_table_name_dict):
@@ -112,3 +113,17 @@ def test_sqlplot_boxplot(ip_with_MSSQL, cell):
     out = ip_with_MSSQL.run_cell(cell)
 
     assert type(out.result).__name__ in {"Axes", "AxesSubplot"}
+
+
+def test_unsupported_function(ip_with_MSSQL, test_table_name_dict):
+    # clean current Axes
+    plt.cla()
+    out = ip_with_MSSQL.run_cell(
+        f"%sqlplot boxplot --table " f"{test_table_name_dict['taxi']} --column x"
+    )
+    assert isinstance(out.error_in_exec, UsageError)
+    assert "Ensure that percentile_disc function is available" in str(out.error_in_exec)
+    assert (
+        "If you need help solving this issue, "
+        "send us a message: https://ploomber.io/community" in str(out.error_in_exec)
+    )
diff --git a/src/tests/test_magic_plot.py b/src/tests/test_magic_plot.py
index 780512092..6e0834af1 100644
--- a/src/tests/test_magic_plot.py
+++ b/src/tests/test_magic_plot.py
@@ -4,6 +4,34 @@
 from IPython.core.error import UsageError
 import matplotlib.pyplot as plt
 
+from sqlalchemy.exc import OperationalError
+
+from sql.plot import _summary_stats
+
+
+def test_summary_stats_success(tmp_empty, ip):
+    Path("data.csv").write_text(
+        """\
+x, y
+0, 0
+1, 1
+2, 2
+5, 7
+9, 9
+"""
+    )
+    ip.run_cell("%sql duckdb://")
+    result = _summary_stats(None, "data.csv", column="x")
+    expected = {"q1": 1.0, "med": 2.0, "q3": 5.0, "mean": 3.4, "N": 5.0}
+    assert result == expected
+
+
+def test_summary_stats_failure(tmp_empty, ip):
+    ip.run_cell("%sql duckdb://")
+    with pytest.raises(OperationalError) as e:
+        _summary_stats(None, "data.csv", column="x")
+    assert 'No files found that match the pattern "data.csv"' in str(e)
+
 
 @pytest.mark.parametrize(
     "cell, error_type, error_message",
@@ -166,3 +194,4 @@ def test_sqlplot(tmp_empty, ip, cell):
     # maptlotlib >= 3.7 has Axes but earlier Python
     # versions are not compatible
     assert type(out.result).__name__ in {"Axes", "AxesSubplot"}
+