I went into a client to discuss some points. While there a representative of a magazine the newsagent handled arrived. Since he knew the person well, he introduced me knowing that I would find it interesting. So during the conversation, I brought up a letter printed in National Newsagent from Peter Webster the editor of F&B.; Much was confirmed although the business of sending it by mail to newsagents was considered to be over the top.
Coming back, I decided to look at one of the complaints by Peter Webster. He states the magazine has to be completed by late May, to be released in late June and no figures of what was sold will be available to the publisher of the magazine till late September. He wants the figures earlier.
These figures they cannot get it out of supermarkets the figure partly as the supermarkets probably will not take them and partly I think only network get the sales data from supermarkets. This to some extent makes them more important to newsagents as if the public want this magazine they have to come to the newsagent.
Now only a small percentage of newsagents are sending sales data. Much of it is subject to major criticisms by the data analysist because of its low quality. Something we have been talking to them at several meeting recently.
So I decided to do some research by doing a simulation. Now say we have out of 3,600 various sized newsagents, 200 newsagents again of various sizes are sending decent data every day.
What would show the sales estimate for a magazine per day based on Xchangeit sales data over 12 years. In the following charts you can see the results for yourself.
The average error plus or minus is in blue.
The Yellow line shows within 20% accuracy
Pink line shows within 10% accuracy, which is fine for an estimate.
Assuming a typical newsagent sells one copy of the magazine a month. This is a questionable magazine to handle anyway but for the sake of completeness I did it.
The average error in the supplied sales data to the publisher was about 33% overall. So the publisher could know here that his sales a month is somewhere between 2520 to 4680. It is certainly is better then getting nothing.
It looks like this over a 12 year period.
Redoing the calculations assuming a typical newsagent sells five copies of the magazine per month
The average error in the supplied sales data was about 13%. As you can see the figures here are quite reasonable. Half the time he is getting figures within 10% of the actuals daily and almost always within 20%.
Assuming a typical newsagent sells ten copies of the magazine per month
Now our average error in the sales data to the publisher was about 9%. You can see how we are okay.
After ten a month it does not matter as we are close to 100% accuracy anyway.
Note I have assumed here that there is no magazine decay (the sales decay is the fall in magazine sales after release date) which would actually improve the figures. Because you can see here as the sales go up the accuracy increases.
As this analysis shows there is no reason with the existing technology and systems in place the industry cannot supply the figures these small magazine publishers require. Why is it not done now because it is hurting the industry?
