Affinity Analysis and the “man aisle”
Part 1 of a 4-part series
Ok, here’s my latest attempt to explore a technical topic. I’m going to stick with what I know or can easily understand, but I’m inviting the real expert to join the discussion to explain the more technical aspects of this.
I recently came across a story about a Manhattan grocery store that introduced a “man aisle” to some fanfare. The idea is simple: guys do more of the family shopping these days so why not make it easy on them? From beer, chips, and salsa to razors, hot sauce, and beef jerky – the man aisle puts all the guy-goodies in a single place for convenient shopping.
To me, this seems like an intuitive approach to affinity analysis – something my expert, Shantanu Goswami at SAP, speaks a great deal about. As he explains it, affinity analysis – also known among retailers as market basket analysis – is basically the attempt to gain insight into which products people buy together.
The chips and salsa scenario is a good example. Let’s say that you notice a correlation between people buying a certain brand of chips with a certain brand of salsa. In such a case you may not want to put both on sale at the same time but rather hope that a promotion on one of them drives sales for the other. Or at the very least, you may want to place each product in close proximity.
I don’t know much about the level of analysis that went into the “man aisle,” but affinity analysis can be serious business for retailers who need to drive sales and profit margin every quarter. The sheer volume of data available for analysis and insight can be quite staggering. A typical data sample for a large chain retailer could run into the millions of records.
Shantanu can share a bit more insight around the technology that can make this happen. Its all about crunching big data in real real-time. The “man aisle” is a great example of affinity analysis. I’m sure there are others and I’d love to hear about it.
I’ll stop here for today, but stay tuned. The next topics for exploration in this series will be the drag along affect, top seller analysis, and hoarding behavior.
Learn more about affinity analysis for retail.
Technically, Affinity Insight consists of very interactive SAP Business Objects Dashboards as a front end, a JAVA middle layer, and a HANA data base which performs most of the statistical computations.
You lost me at JAVA middle layer. Could you provide a real-world example from your experience that demonstrates affinity insight?
Affinity Insight is an SAP HANA based data mining tool specifically for retailer’s point-of-sales data. It provides statistical functionality and enables users to perform fast and extremely flexible analysis of market basket compositions across time, store and product hierarchies. I
For example - if a customer buys a digital camera, how much will he/she spend on batteries, camera bags or other equipment? That's darg-along effect. More on that in the next blog ... stay tuned.
Thanks for the follow up Shantanu. Looking forward to your blog!