Last Friday, Palantir raised $450 Million in a new round of funding, at a valuation of almost $20 Billion, making it the fourth most valued “startup” to date with almost 1 Billion in funding including Founders Fund, Tiger Global Management, and In-Q-Tel, the CIA’s investment arm.

But it’s not just big data that generates big money (for the software provider) and big value (for the organization that has [access to] it). It’s big analytic power. And, as SI has indicated repeatedly, the data set doesn’t necessarily need to be that big to identify considerable savings opportunities.

A million transactions might not be more insightful than 1,200 transactions. If the transactions are for 10 different products from 10 different suppliers over the course of the year, a single summary transaction for each month for each supplier-product pair that summarizes the lowest price paid, the average price paid, the highest price paid, and the total paid is just as informative from a spend analysis perspective. Given this data, the buyer can see, for each product, how much money it would have saved if it always bought at the lowest price, how the price is trending, and how much could be saved by using a contract to lock the product in at a price less than the current market price. The other 998,800 transactions are not needed.

In other words while you need large spend cubes to find value opportunities, which will often depend on redefining categories, redefining shipping lanes, redefining delivery schedules, and so on, you can often get away with cubes that are at most, hundreds of thousands of well defined (summary) transactions (for the right time period). Millions of transactions are typically not necessary, and that’s why you can do enterprise wide spend analysis on a laptop with the right spend analysis tool (like busiq.com) as you can generally define a transaction set of just a few million transactions that covers the last three years and fits in memory!