Sales data Analytics

At first glance you think that total analytics you can get from third party vendors by just including the Java script, then why to have sales data analytics.  The reason is there is lot more data  which is not visible/present on the website, so they cannot simply have analytics of the data which they cannot see. Example: Customer returns data, category data, shipment data, customer demographics data. Capturing all these analytics  from web analytics may not be possible depending on your customization and granularity and data availability on the website. We will see some example analytics.

• Customer demographic data
• At what time men are ordering the items (lunch time?)
• At what time women are ordering the items?
• Metros and big cities women are interested in which items? and when they are purchasing?

•  Order data
•  How many items are getting returned per month, per category
• What is the average order size.
• At what time you are getting highest number of orders ( It is your peek time, monitor everything)
• From which geographical location, you are getting highest orders ( helpful for marketing)
• Which category items being returned most (may be some thing wrong with listing or packing)
• To which area highest number of shipments are happening (so that you can setup your next warehouse there)

•  Inventory data
• Which category is moving fast
• Which products are procured most even when customer is waiting for that.
• What are best selling products (so that to see on website)
•  For which items customer asked to Notify highest number of times.

When to have such system?

You can get very good analytics for some time by Google Analytics or any other tool. But when you are expanding after certain time, after few months or years, you can plan to have in house analytics on sales data. 

How to pull the data?

Definitely you cannot run on the production database, it will hamper the production system. Best thing would be to work on secondary database which is in sync with primary database. 

1. Set up a secondary database which is in sync with the primary
2. Copy the complete data from secondary database data on your local/ temporary database - call it as dump
3. Run queries on the dump and extract the results you want.
4. On the long run you cannot manage all of this, so you have to eventually have to take help of the too good third party tools like Pentaho or Informatica. This involves writing ETL's