The Magazine Publishers of Australia selected the following company Boston Analytics to provide an independent and robust evaluation on their pilot magazine tests to the ACCC. The public release of this report by is expected in a few days see here.
We have worked closely with Boston Analytics, to help them, however we were not given the data, we requested to do our study what we could do the analytics for our clients. So although we have fewer shops to compare then Boston Analytics, we do feel we had access to more non magazine shop data, plus we used a longer and more extensive shop history, and we think our data mining analytic tools are better.
Data mining (smart software) is something only Pos Solutions retailers have in our market space. If someone in our market space disagrees with this statement, show me your shop reports with it absolute, square, R squared, correlations, squashed, interquartile, etc. errors all documented for equation generations, etc. Data mining in retail is a field that we consider ourselves, a market leader.
With our data mining, we compared magazine sales by unit price, number and profit over five(5) years with every factor we could think of and ended out with the following list of factors to study.
Trial of magazine, Plain packing of tobacco, Hidding the tobacco, Julian date, Summer, Autumn, Winter, Spring, Monday, Tuesday, Wednesday, Thursday, Friday, Saturday, Sunday, Payout_number, Payout_unit, Payout_profit, Others_number, Others_unit, Others_profit, Card_number, Card_unit, Card_profit, Delivery_number, Delivery_unit, Delivery_profit, Drink_number, Drink_unit, Drink_profit, Eftpos_number, Eftpos_unit, Eftpos_profit, Fax_number, Fax_unit, Fax_profit, Gift_number, Gift_unit, Gift_profit, Instantpay_number, Instantpay_unit, Instantpay_profit, Instant_number, instant_unit, Instant_profit, Lollies_number, Lollies_unit, Lollies_profit, Lotto_number, Lotto_unit, Lottopay_number, Lottopay_unit, Lottopay_profit, Misc_number, Misc_unit, Misc_profit, Paper_number, Paper_unit, Paper_profit, Phonecard_number, Phone_unit, Phonecard_profit, Photocopy_number, Photocopy_unit, Photocopy_profit, Stationery_number, Stationery_unit, Stationery_profit, Tobacco_number, Tobacco_unit and Tobacco_profit
I then did almost eight(8) million calculations in our data mining and over 500 billion evaluations on each of the stores, I checked. I know it's a lot of work which is why few do it.
Factors that were shown to be important in magazine profitability were days of the week, card sales, the amount and number of instant pay outs as people who win in often buy magazines, newspaper sales, drink sales as people that are thirsty come into the shop to get a cold drink and while they do this sometimes buy more and phone card sales for similar reasons as drinks.
However, my conclusion is that the trials had no effect on magazine sales.
This conclusion is not surprising to me as from the sales view it made no difference if the non sellable magazines are culled by the newsagent before they go on the shelf, or they are culled by the magazine distributor before they are sent. Although from an economic view, I think it saves a lot in processing handling, bookkeeping, transport costs, etc. if the magazine distributor does the cull.
So in summary I feel that the magazine trials show that the publishers can, with no loss of sales can do a better job of distributing magazines.
Interestingly those on the trial that I talked to, felt they needed early returns as some of the magazines they got they did not like and also they needed the ability to change orders as some magazines that were culled they felt they could have sold.
I look forward to reading Boston Analytics take on the matter.
