"Life is a voyage of discovery" -
Arnold Toynbee (
apparently)
Frankly, I would rather today had been less of a voyage and more of a cruise, but having cast off from my safe harbour this morning and fought through the stormy waters of web site analytics, I thought I would share my discovery with you.
My aim was to enable full post feeds for Roland (who
requested it) and for the rest of you (who did not, but will probably find reading my stuff easier in the long run because of Roland). Previously I had only been issuing partial feeds in order to monitor the articles that people wanted to read, so that I can write more about the stuff people are interested in and less about the stuff you aren't. I figured if you read the clip and then clicked through, I was onto a winner.
So I signed up to
Feedburner. If I'd been a greenfield site, it would have been fine. The
initial setup was easy: just enter the URL and select the feed type (ATOM or RSS), hit burn, install a "chicklet" and you're away, right? Well, not quite... (and how did an
icon to indicate the availability of an RSS feed get named after a small chicken anyway?)
The problem was that I already had existing feed subscribers and I needed to get them through Feedburner too if there was to be any point in using the service. Legacy customers aren't just a problem for service providers, y'see. Even bloggers find it difficult to "migrate users onto a new platform" and "benefit from a single infrastructure".
This is no criticism of Feedburner and their support pages are so good that they are practically useless. There is so much information on there (34,286 posts) that you can't find anything without working really hard for it. A search for "feed file size" gives you 1,708 answers, while "blogger atom" gives you 1,853 matches - in chronological order based on the last post . All this is written for (and by) people with so many different levels of technical ability, that I got lost. Hopelessly lost - I'm not a web developer, I never was and I never wanted to be, but now running this blog has forced me to confront my demons.
I thought that blogs were supposed to be easy - making publishing on the web available to the masses...
So, here is what I learned. It is all written up
here and
here; call me a numpty if you like, but I found it really hard to wade through all the different explanations at varying technical levels for people on different blog providers so I am (hopefully) simplifying it below. If just one person saves half the amount of time I spent figuring this out, then it will have been worthwhile. The rest of you can have a good laugh at my expense over your Monday morning coffee. How did he not know that?!?
Firstly you need to start by changing your blog setting so that instead of publishing the file that everyone has subscribed to (in my case atom.xml), you publish your feeds as something else (eg atom2.xml).
Secondly, you need to tell Feedburner that instead of looking for atom.xml, it should pick up atom2.xml.
Thirdly, and this was the scary bit, you had to create (or modify if you have one already) the .htaccess file to include the following lines:
Redirect 301 /atom.xml
http://feeds.feedburner.com/ipdevRedirect 301 /rss.xml
http://feeds.feedburner.com/ipdev...and upload them to your hosting site in the root directory (htdocs, httpdocs) of whatever domain, or in my case subdomain, you are using. A good resource I found is
here.
Oh, yes. That's the other thing I learned. If you are using a blogspot hosted service, you're stuffed. You can't change this .htaccess file unless it's yours and yours alone. And, you have to make sure it's uploaded as an ASCII file and not a binary. I lucked out there - I have no clue what I would have done if it hadn't defaulted to ASCII. Cried, probably.
Then, scariest of all, I had to delete the old atom.xml and rss.xml files off my FTP server. By this stage, I was really crossing my fingers and hoping that I hadn't screwed up somewhere along the line, because if I had, there was no way I could have unravelled all of this and gone back to partial site feeds.
Once I had done all of this, I found I had more to do. There is such a thing as
autodiscovery which enables feeds to be found if you know the blog's web address. In order to change this, I had to hack around the way the blogger creates this part of the page. Oh yes, and it's different if you have "new" blogger and "old" blogger, which was another thing that confused me. It turns out I'm on "old" blogger, even though I signed up 2 months ago. How do you know which version you are on? Look
here... it's obvious - as is everything if you know where to look.
Anyway, for those who are trying to follow me here, this is what I did. Go to your blog, right click, view source, find the bit that matches the code in this
link, copy it, go to your blogger template, delete <$BlogMetaData$> and paste instead the copied bit from the source file.
My source file had two lines, one for atom.xml and one for rss.xml and I spent ages trying to figure out what to do with the rss bit. You see, when you set up feedburner, it asks you which one to look for, I chose the atom feed, so I knew I had to replace the
http://blog.ipdev.net/atom.xml with
http://feeds.feedburner.com/ipdev because it said so in the
instructions (see, I do read them occasionally!).
Another tidbit. I found a "
help article" saying "We do not process source feeds that are larger than 512K. Frankly, there's no good reason for a feed to be quite this large". Fair enough I thought, and the piece goes onto to explain how to set the number of posts visible on the home page in blogger. The problem was that my feed still had 25 articles in it, even though the home page has two. I think this is a feature of atom, but seeing as my 25 articles were about 280k, I didn't care any more.
In my voyage of discovery I also came across some very interesting stuff, which isn't directly relevant here, but is worthwhile linking to anyway. Check out this
extremely thorough review of various web site analytics providers. The one I like, but just can't justify the $15 for is
VisitorVille, which strikes me as cool and nuts in about equal measure. Anyway, I've had StatCounter for a while, I have now installed the Feedburner (feed + site) analysis tools and Google Analytics, so one day I'll have to write up what I found in a back to back review at some point.
So why did I set about on my voyage this morning? The answer is simple - to give my customers what they want (while making sure that I can still count the number of people interested in what I write).
Why am I writing about this? To help people in the same situation with Feedburner, but also to get you thinking about service migration. It is not good enough to write a few basic FAQs and deal with the problems on the fly or you end up where Feedburner is now - with a service that is great for greenfields, but does not do well when people are switching an existing service to you and want to keep elements of what they have been using before.
Getting people to move their services over to you is the only way to beat churn and grow in a saturated market. We talk about "lock-in" and "stickiness" so we are well aware of efforts by incumbents to keep customers, but surely there is more that can be done to help customers escape the quagmire and "seamlessly migrate" to you? Isn't there?
Labels: blog analytics, blogger, feedburner, service migration
# posted by Jeremy Penston @ 5/05/2007 12:07:00 AM
