Start with a data dump from your AP system. For illustration purposes, imagine we have a million rows of data.

First, categorise every row as “Other”. (This allows you to claim that you have categorised 100% of your data and you’re only on step 1!)

Second, categorise each row according to a GL/Category mapping table. (This ensures you get the tail spend.)

Third, categorise each row according to a vendor mapping table for your significant vendors. (This ensures you’re not embarrassed by a PA who has for reasons unknown placed a KPMG invoice into “Kitchen Supplies”.)

Fourth, categorise each row according to a mapping table for vendor and GL. (This ensures that you split your Staples spend into stationery and office equipment etc.)

And finally, categorise each row according to key phrases in the description. (This ensures you pull out PWC legal spend from the Audit category.)

But today, I played around with Splunk. This is not as dirty as it sounds. Splunk is an IT log analysis tool created to assist IT Helpdesk staff diagnose faults by sifting through terabytes of IT log files. It is a refreshingly different view on spend data. Useful? My jury is still out. Fun? Embarrassingly, yes.