diff --git a/python/sql/messy_data/__init__.py b/python/sql/messy_data/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/python/sql/messy_data/messy_data.py b/python/sql/messy_data/messy_data.py
new file mode 100644
index 0000000..541aaf3
--- /dev/null
+++ b/python/sql/messy_data/messy_data.py
@@ -0,0 +1,47 @@
+class MessyData():
+    """
+    ## Messy Data
+    Your company has an internal policy to determine your customers' credit limit, but this procedure has been questioned recently 
+    by the board as being too conservative.
+    Your CEO wants to increase the current customer base credit limits in order to upsell a new line of products. 
+    In order to do that, the company hired several external consultancies to produce new credit limit estimates.
+    The problem is that each agency has produced the report in its own format. Some use the format "First-name Last-name" 
+    to identify a person, others use the format "Last-name, First-name". There is also no consensus on how to capitalize each word, 
+    so some used all uppercase, others used all lowercase, and some used mixed-case. Internally, the data is structured as follows:
+
+    Table: customers
+    ================
+    id: INT
+    first_name: TEXT
+    last_name: TEXT
+    credit_limit: FLOAT
+
+
+    Table: prospects
+    ================
+    full_name: TEXT
+    credit_limit: FLOAT
+
+
+    Keep in mind that the agencies had access only to a partial customer base. There is also the possibility of more than one agency 
+    prospecting the same customer, so it's highly likely that there will be duplicates. Finally, they've prospected customers that 
+    were not in your customer base as well.
+    For this task you are interested in the prospected customers that are already in your customer base and the prospected credit limit 
+    is higher than your internal estimate. When more than one agency prospected the same customer, chose the highest estimate.
+    
+    You have to produce a report with the following fields:
+    
+    first_name
+    last_name
+    old_limit [the current credit_limit]
+    new_limit [the highest credit_limit found]
+
+    In order to solve this exercise you may pick your preference / combination of:
+        - Python and/or Pandas
+        - SQL or equivalent DSL language
+        - an ETL tool
+
+    """
+
+    def report(self):
+        raise ValueError("Not Implemented.")
\ No newline at end of file
diff --git a/python/sql/messy_data/test_messy_data.py b/python/sql/messy_data/test_messy_data.py
new file mode 100644
index 0000000..c2b2ba4
--- /dev/null
+++ b/python/sql/messy_data/test_messy_data.py
@@ -0,0 +1,8 @@
+import unittest
+
+from .messy_data import MessyData
+
+class TestMessyData(unittest.TestCase):
+
+    def test_report(self):
+        pass