diff --git a/DATA.md b/DATA.md
new file mode 100644
index 00000000..bf1fd05c
--- /dev/null
+++ b/DATA.md
@@ -0,0 +1,44 @@
+### questions.xlsx
+
+This file contains definitions of the questions. It should have one tab
+per section with all the questions for that section one per row.
+
+The columns are:
+
+ * question_no - either a number of number followed by letter for multi
+     part e,g 1a
+ * topic
+ * question - question text
+ * criteria
+ * clarifications
+ * how_marked - one of Volunteer Research, National Data, FOI
+ * weighting - unweighted, low, medium, high
+ * district - Yes if applies to district councils, blank otherwise
+ * single_tier
+ * county
+ * northern_ireland
+ * question_type - how the question will be answers (see below)
+ * points - number of points question is worth before weighting
+ * option_[1-n] - possible option answers for multuple choice questions
+
+#### Question types
+
+ * Y/N - true/false question
+ * blank - true/false
+ * Tick all that apply - multi select question
+ * Tiered answer/multiple choice - multiple choice question
+
+### volunteers.xlsx
+
+This is for bulk assigning volunteers to the first mark. The columns
+used are:
+
+ * First name
+ * Last name
+ * Email
+ * Council Area - volunteers own council which they will not be assigned
+ * Type of Volunteering - used to determine how many councils to assign
+ * Assigned Section - section to assign user to
+
+You can have multiple tabs for different stages but the name of the used
+tab is hard coded in the script.
diff --git a/README.md b/README.md
index aa8232a8..3c331fcd 100644
--- a/README.md
+++ b/README.md
@@ -48,6 +48,8 @@ questions and assign them to the relevant councils. The questions data
 should be downloaded as an Excel sheet from the Google sheet populated
 by CEUK.
 
+Details on file strucures can be found in DATA.md
+
 ### Data structure
 
 See the separate ARCHITECTURE.md file for full details on the underlying
@@ -100,7 +102,11 @@ Once the user has been set up they can set a password by visiting the
 django password reset page.
 
 For initial volunteer setup there is an `import_volunteers` management
-command.
+command. This will take a list of volunteers and assign each of them a
+set number of councils in the section they are assigned until the
+councils have run out. If there are more volunteers than assignments
+then some volunteers will not be assigned. The number of councils to
+assign is currently hard coded in the script.
 
 ### Exporting marks