MPA categorization of magazines

I was asked today just how big a job is categorisation of magazines such as what the MPA and Bauer Media were talking about recently at the ACCC conference. The answer is, its big.

In Australia a small reseller of magazines per year goes though between 1500 to 2000 titles. Title Tracker when I saw it last had a very extensive list which had about 5000 titles available in Australia broken down into more than 50 categories. It also had a section on special interests.

If the MPA were to redo it as stated they would they are looking at a couple of man-months of work. That is before they can even start the trial. If they do not do this then no retailer is better off then we are now.

That does not include the demographics of the shops as they stated and the analytics.

The demographics of a shop are not easy or particularly accurate to determine. They, firstly, need to identify the feeder area for the shop. Afterwards check through the census reports to see what demographics this area has. Then they need to determine the traffic passing though. Trains, trams and freeways give little for marketers while a busy main roads often give much. They also need to check who works in the area, for example a large office block nearby can change the demographics of a small shop. It takes a townplanner quite a while to do this. You also need to check the competition because if there are two or three shops in the area you need to determine what part of the market the one you are measuring has.

A typical analytic project like this one often the analysts need a month just to get started. You tend to do run after run after run until you start getting something that makes sense. It's often a case of exhaustion. We have done enough let's just proceed with what we have.

Here what you would need to do is see what retail customers (demographics) have bought based on what actions they have taken previously in this case using the data from XChangeIT. You need a lot of skill as its quite complex. People that do this, earn big bucks as to do, for example, a statistical customer segmentation to understand retail customer needs and their intentions is not easy. Then they need to determine the optimal number of magazine segments and do what we call a market basket analysis to identify the cross sells. Sometimes you suggest people move items from one category to another simply because it sells better in another category. You then have the start of your categorization, which brings us to my earlier point.

An interesting point is that Fairfax did a smaller but in many ways a similar study to this many years ago and found that the retailer was more accurate then their analysts. The retailer was accurate to 5%, the analysts to 10%.

I am willing to keep an open mind, to be fair but somehow I doubt this will be done.