Dangerous Precedent

People looked at me a tad funny when I said I was joining Wired UK’s launch team. Print magazines are a dying medium, they said. What the hell are you doing?

Actually, we’ve been doing very well. Anecdotally, we’ve been selling out at quite a few of our stockists. But that, right there, is the problem: for someone like me, with a background in reader statistics-heavy online media, it’s been frustrating how little we know about our readers. While a website can tell who its readers are, where they come from, when they get there, how long they spend where, and where they go to when they leave, all within milliseconds, all a magazine knows is how many it sold – and not necessarily down to the individual shop – and all within a few months of the magazine first going on sale. That’s it.

This has three immediate effects. Firstly, the demographics that magazines cater too are nothing more than educated guesses. Whether this is a good idea or not is another post, really: I’m not at all convinced that it’s a bad idea to have to invent your reader rather than chase real peoples’ opinions into a spiral. (I’m not going to comment on what this means to the advertising industry: they know all this already.)

But second, we can’t properly optimise the distribution. All magazines sell out in some places and don’t sell at all in others – that’s understandable – but currently there’s nothing whatsoever that a publisher can do about it in anything close to real-time. We might look at maps of sales across the country and change the number of copies sent to different shops, but this happens with a lag of, again, months. Even the most efficiently distributed magazine still pulps thousand of copies every month after they were left unsold somewhere – even when the same magazine leaves readers disappointed because it was sold out in their local newsagent.

Both of these points bring me to my first point, which is this. Sometimes in an assailed content business, it’s not the content that’s breaking: it might be the business bit. Reinventing the content at the same time isn’t necessarily the right thing to do: but you’ll never know this unless you actually get into the business itself. Being a consumer isn’t good enough. I’ll be talking about this more, I think, but not today.

Anyway, the third upshot of this sort of thing is that it does start to present opportunities, once you manage to get the data. And get some data I did. I managed to get a spreadsheet of our stockists, and have made a prototype webapp thing for the iPhone3 to show you the places to buy Wired UK nearest to your current location. It won’t tell you if they have any left in stock, but it will tell you who, at least, had some in the first place.

To try out this beta application, point your iPhone (upgraded to version 3) at:

benhammersley.com/whereswired

It’s very rough, and slow, and the data isn’t complete – it’s missing quite a few shops at the moment – but the basic idea is there. email me if you’ve got any feedback – it’d be greatly appreciated.

Hereafter the technical bits.

The address data came in an Excel spreadsheet, which I exported as CSV. I then wrote some Perl to go through that line by line, and use the Google Maps Geocoding API to convert the addresses to Lat/Long. This is all written out into an XML file (it’s quite verbose because I only wanted to do this once, and use the data for other, non-public, apps). The app itself simple JavaScript Google Maps API, that uses two cunning tricks: this Clustering code, and this code that uses the new nascent W3C geolocation standard that the iPhone3 supports, like so:

navigator.geolocation.getCurrentPosition(foundLocation, noLocation);

function foundLocation(position)
{
  var lat = position.coords.latitude;
  var long = position.coords.longitude;
  alert('Found location: ' + lat + ', ' + long);
}
function noLocation()
{
  alert('Could not find location');
}

It’s not perfect, as I say. The app loads all of the country’s shops in one go, which is data-heavy to say the least, which makes it slow. It doesn’t show an icon where it thinks you are. It needs information available for each pointer. And so on: email me. I’m doing more work on it later in the week.

#