diff --git a/How-fast-are-really-the-analytical-DBMS%3F.md b/How-fast-are-really-the-analytical-DBMS%3F.md
index dc8e2f2..cbe7f50 100644
--- a/How-fast-are-really-the-analytical-DBMS%3F.md
+++ b/How-fast-are-really-the-analytical-DBMS%3F.md
@@ -10,6 +10,32 @@ I did a small experiment with a synthetic dataset I created for the PyCon Spain
 
 ![Table](https://git.guillemborrell.es/guillem/blog/raw/branch/main/images/analyticdbms/Screenshot_20230212_215503_Chrome.jpg)
 
+The query is designed to crash naive query engines, with a CTE and two nested aggregations:
+
+```sql
+with by_nplayers as (
+SELECT 
+    max(toss) as toss,
+    count(*) as qty,
+    count(*) as nplayers,
+    game
+FROM 
+    boards
+GROUP BY
+    game
+)
+select
+   sum(qty) as qty,
+   toss,
+   nplayers
+from 
+   by_nplayers
+group by
+   toss, nplayers
+order by
+   nplayers, toss desc
+```
+
 Here are the results for postgresql
 
 ![Postgresql timings](https://git.guillemborrell.es/guillem/blog/raw/branch/main/images/analyticdbms/Screenshot_20230212_213038_Chrome.jpg)