EgoReFlogging: Maximum PC

Ok checking through my logs for the first time in ages – have to say thanks to the people over at Maximum PC for featuring me in their fave podcasts recently! http://www.maximumpc.com/how_to/reprint_2005-03-15.html 😀

Also I need to recheck my logs because this *couldn’t be right* – I can live with 21,586 visitors to MutantPop in the last week – scary thought but MutantPop has many parts – but it’s telling me I have 13,695 visitors to my feed and my top browser is iPodder:

1 iPodder/1.1.4 +http://ipodder.sf.net/ Hits: 3,304 Visitors: 2,726 12.89%
2 iPodder/2.0 +http://ipodder.sf.net/ Hits: 3,207 Visitors: 2,324 10.99%
3 iPodderX/2.2.8 (http://iPodderX.com) Hits: 2,381 Visitors: 2,160 10.21%

If that (+ all the other RSS/podcasting clients, from Nimiq, Dopple, NNW, Bloglines etc, I haven’t counted the ones I haven’t heard of) is correct I can’t have nearly 12,000 subscribers? I assume my log is seeing them as hits.

Still 12,000 hits just from Podcasting/RSS clients in a week? 😮 …but if everyone has their clients set to check every 30 minutes (worst case scenario – I guess actually most people have the default set which for iPodder is 2 or 4 hours) it’s 139 people. Phew.

Anyone know of more reliable ways to analyse these logs (or check for unique visitors) cos I’m sure I don’t have 12,000 subscribers…or do I? =:-o

9 Comments

  1. March 29
    Reply

    Your math is making me dizzy. in bloglines the subscibed number says something like 28. Which is practical terms usually represents 1/6 of how many subscribers you really have. Then again, I invented that in my head. Along with my large collection of friends.

  2. March 29
    Reply

    I think I’m right – logs are meant for websites not Podcasts; most log tools I guess count a different visitor each time the same iPodder accesses the feed as it’s more than 10 or 20 minutes and that’s longer than most people access one page so it thinks it’s someone else. If I buy a nice shiny new log tool it’ll tell me ‘authenticated users’ – ie. IP addresses and that’d be more useful.

    Interesting – how would bloglines know how many people are hitting my feed as it’s on my server (?). I assume that’s people using bloglines to access Radio Clash, and that’s about right from my logs.

    I get around 100-150 requests (or pings?) via the forums and people clicking on the blog per show, so I’d guess around 150-200+ via iPodder would be about right…

  3. March 30
    Reply

    I’m not sure how your log analyizer determines visits, but you’re probably right that it doesn’t represent unique visitors. I’ve been using FeedBurner to host my podcast feed and have spent quite a bit of time looking at the statistics they provide, which are quite nice (I’m a numbers guy you see). They seem to quantify a feed’s circulation by unique IP and user-agent pairs. This is going to miss multiples of the same version clients behind a common firewall, but will also overcount people who hit the same feed with multiple clients. I think its a good estimate, but have done some checking to validate it.

    I’ll try not to make anyone dizzy here, but this’s what I’ve seen from my FeedBurner stats. Podcatching clients tend to hit a feed anywhere between 6-10 times/day (8 hits/day on average). So a quick and dirty way to estimate your circulation is to divide the average number of hits/day on your feed by 8. I’ve compared circulation calculated this way with real downloads. This is where I get really geeky. Importing my log file into a spreadsheet, I take the total kBytes transfered for each show’s mp3 and divide by the mp3 file size, which gives me an approximate number of downloads for each show. The total number of downloads is typically 10-20% higher than the FeedBurner reported circulation. This could be due to errors in either calculation or people downloading the mp3 directly from my blog.

    Unfortunately I haven’t found a log analyzer that will do this all itself. Please let me know if you do.

  4. March 30
    Reply

    Well, my iPodder is set for once every 6 hours… so you can take that into account. The other sure fire way to determine actual downloads is to look at your bandwidth… although that doesn’t really help determine the who’s and when’s… My new Podcast hosting people LIBSYN actually track the number of unique ISP downloads per podcast ) according to client. Although, strangely, I’ve noticed my downloads from iPodders grows throughout the month… no idea why. they start at half their normal downloads and slowly grow.

    We really need a Podcast Stats application that can show us these kinds of stats from our own servers, without painful geekery.

  5. March 30
    Reply

    thanks for that Devan – sadly I can’t get logs for the MP3 files themselves, the webspace won’t allow scripts (hence the redirect I coded as a simple ‘counter’).

    I’ll try that and see how I go…oh for a fully featured open source log tool that doesn’t require you to faff around with CLIs or CGIs. I’m using Weblog Expert Lite, full version is great but I can’t sanction paying nearly 50 quid for it…

  6. MK
    March 31
    Reply

    Congratulations. You’re now officially bigger than the Dawn and Drew show. 😉

    MK

  7. MK
    March 31
    Reply

    Ben, I’ve always suspected that a subscriber download counts for 2 downloads in Libsyn for some reason. One subscriber download and one non-subscriber download. I still think the best metric is unique IP address……

  8. March 31
    Reply

    No I’m not as big as D&D – no way!

    So using Devan’s formula I think I have abour 200 subscribers + 150-200 people (definitely) clicking from forums – whether they fully d/load I don’t know…so 350-400 I’m really happy with 😀

  9. John
    April 1
    Reply

    I can’t imagine you only have 200 subscribers. Hell, I’ve discovered your podcast here in Halifax, Nova Scotia, and I’m a relative podcast newbie. You’ve got a great show. I’ll bet substantially more than 200 folks are subscribed.

    Regardless, keep Radio Clash coming. Wonderful stuff.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.