Occam's Razor by Avinash Kaushik

Top Ten Web Analytics Blogs: July 2007

34 User Comments

1. WebAnalytics.be 5ème dans le top 10 des blogs WA « Negligible quantities Says:
   July 27th, 2007 at 02:57

   [...] WebAnalytics.be 5ème dans le top 10 des blogs WA Published July 27th, 2007 Web Analytics Avinash Kaushik vient de sortir son classement des 10 meilleurs blogs sur le web analytique. Il avait commencé cet exercice en juillet 2006 et les résultats sont intéressants: [...]

2. Anil V Umachigi Says:
   July 27th, 2007 at 06:58

   Hey Avinash!

   Great Post! I'll have to include few of those blogs in my blog roll! and subscribe for the feeds too.

   Thanks
   Anil
   
   Web Analytics Pulse

3. Eric T. Peterson Says:
   July 27th, 2007 at 07:43

   Avinash,

   Great list as usual and thanks so much for honoring my request to be dropped from the system. I'm sure you'll agree that making a slot for Craig (or any of the bloggers in your list above) is well worth it!

   One question: Given the problems that Technorati has with ranking (see my recent post at blog.webanalyticsdemystified.com/weblog/2007/07/technorati-is-a-poor-source-of-blog-ranking-data.html) why not just use the Feedburner subscriber data?

   Or, put another way, can you tell us what happens to your list if you ** only ** base it on Feedburner data? Does it stay exactly the same or does it change somehow? Seems like if you get the same ranking it would make sense to drop the Technorati data that you yourself are describing as having "issues" …

   Anyway, brilliant work as usual and it's great to see that you're still #1!

   Sincerely,

   Eric Peterson
   http://www.webanalyticsdemystified.com

4. Peter W Says:
   July 27th, 2007 at 07:54

   Avinash: This is a great improvement to the ranking system! What I like the most is non linear regression part, it makes it harder for the established bloggers while making it easier for new bloggers to climb up the ranking.

   It is nice to see positive energy on the blog (another thing that you have in common with Kevin).

5. Nathan Smith Says:
   July 27th, 2007 at 08:02

   Avinash I am one of your newer fans, thanks for the inspirational speech at shop.org. I have just started reading blogs and have not added all the blogs on your list to my newly installed feedreader.

   It was great to meet you in San Diego.

   Nate

6. Nate Sidmore Says:
   July 27th, 2007 at 08:46

   Interesting points in your blog post Eric (blog.webanalyticsdemystified.com/weblog/). Avinash, I'd be interested to see if there were a more accurate way of ranking too. Still, I think it's great to be pointed toward the general area of top Web Analytics blogs.

   Thanks for the service.

   Nate Sidmore

7. Eric T. Peterson Says:
   July 27th, 2007 at 08:53

   Nate,

   Thanks! I should have linked the post in my last comment (but I had not yet had any coffee!) Here is the link to my assessment of why Technorati is a lousy source for blog ranking data.

   And I fully agree with you, Avinash does the entire community an excellent service by tracking great new blogs. That's why I asked to be removed from his list — I am hardly a new blog in this sector and am more excited to learn about new voices. But yeah, since we all work in web analytics, it seems like "data accuracy" is just one of those things we should be working towards.

   Thanks,

   E.

8. Avinash Kaushik Says:
   July 27th, 2007 at 10:23

   Eric : Blogs are a social medium and in as much influence is a key measure of success for bloggers.

   If a blogger chooses to post non-value added content to their blogs then no matter how great they are they will have little "influence".

   Technorati is currently the best option to measure "influence". Just because it is wrong 5 or 10% of the time is not enough to throw the baby out with the bath water.

   After all the tools to measure web analytics that are being implemented by your consulting company on your client websites are wrong by 20% to 2,000% of the time (based on your own blog posts). Yet you, and I as well, continue to use web analytics to make decisions.

   Here is my overall point on data quality on the web: Data Quality Sucks, Let’s Just Get Over It.

   While the quest for perfection is futile it is possible to improve one's ability to make decisions while using imperfect data.

   The addition of feedburner subscribers is my attempt to do that.

   Try this for your blog: Keep your feed subscribers the same, reduce your technorati rank by 25% (well over the approx 5% to 10% "pollution" in technorati data). You'll notice that your rank in the Top 10 will not change that much. The new algorithm helps me accommodate for imperfection (rather well I have to admit, thanks to Kevin!)

   Is this the best way to rank blogs? No. But it is better and while doing this for fun on blogs, or in real life businesses, we learn, we get better, we learn some more, we get better some more. A significantly better strategy.

   Now to your questions……

   Or, put another way, can you tell us what happens to your list if you ** only ** base it on Feedburner data? Does it stay exactly the same or does it change somehow?

   The listing will change slightly, two blogs would show up and the ranking on the bottom half would shift slightly (Craig and Robbin would move up for example).

   But it would not measure "influence" and hence be less optimal.

   Great list as usual and thanks so much for honoring my request to be dropped from the system. I’m sure you’ll agree that making a slot for Craig (or any of the bloggers in your list above) is well worth it!

   While your generosity is appreciated I think it is presumptuous to assume that your generosity "made room" for anyone.

   Using the old system of technorati your WAD blog would have ranked #9 on the top blogs rank. Hence there is a equal chance that:

   * Your feedburner subscribers number would not been enough for you to be ranked just by itself.

   * The combination of feed subs AND technorati ranking would not have been enough for you to "beat" Craig (or anyone else for that matter).

   * Your feed number would have ensured you were on the list and you are indeed being magnanimous.

   I am unwilling to guess which of the above three would have been the case, but it is equally imprudent to assume that one of the above is most definitely the case with missing data.

   Makes sense?

   Industry thought leaders and gurus such as yourself help me push the boundaries of thought, in as much I am deeply appreciative of your feedback and guidance. Thank you.

   -Avinash.

9. Bhupendra Says:
   July 27th, 2007 at 10:50

   Nicely written.

   Thank you so much for providing the scoring worksheet. It is extremely useful.

   Wouldn't the number of daily visitors useful for ranking? I am sure there is some important reason behind not including it. I will be happy if you tell me.

   Thanks,

   Bhupendra

10. Eric T. Peterson Says:
    July 27th, 2007 at 10:50

    Avinash, excellent feedback, and nice comment about my assuming I would make your list using the old method. How presumptuous of me!

    But I guess that was part of what I was commenting about in my post last night, that your system doesn't really have any way to score any blog that is broken up over multiple URLs in Technorati, each of which continue to appear active in their ranking system (see the first half of my post if you haven't had a chance to read it yet!)

    Like I said, it's a Technorati issue.

    I do appreciate that you're actively working to improve the quality of your ranking — that shows real commitment on your part and you're to be commended for that. And I do get that, as one commenter on my blog noted, there isn't much of a better source of data out there … but that's why I asked about just using the Feedburner data.

    Plus, I do agree, your ranking system is fun for everyone. Anyway, congrats again on the new methodology and thanks again for all your hard work. The community certainly does appreciate it!

    Eric T. Peterson
    http://www.webanalyticsdemystified.com

11. Avinash Kaushik Says:
    July 27th, 2007 at 11:34

    Bhupendra : Not using the daily visitor based on my analysis of data from many different blogs. Here were some of the issues:

    1) There is no standard way to collect and process visitor data. Different tools, different methodologies, different numbers. It would be hard to compare apples to apples across blogs.

    2) The visitor number can be influenced by doing short term things. You write something nasty about another blogger / company and you get short term traffic spike but none of those people will stay. Ditto if you do something temporary like get on the home page of digg (some of my posts have done that and its just a temporary spike).

    I have noticed that feed subscribers avoids both of these problems.

    For better or for worse everyone uses feedburner, so one tool and one methodology.

    It is really hard to get short term spikes on your feed subscribers. If you consistently write valuable unique content then your visitors will take the extra step of signing up for your feed becuase they like your content and want it every time you publish it.

    Its that extra commitment that makes feed subscribers such a unique metric.

    So Technorati is a great barometer of your "influence" (even if it is 10% imperfect), and Feed Subscribers are a great barometer of your "content value".

    Does this make sense? Please let me know your thoughts, it will be helpful in the next update of the ranking algorithm!

    -Avinash.

12. WebMetricsGuru Says:
    July 27th, 2007 at 12:45

    Top Web Analytics Blogs – Avinash Kaushik…

    Managed to stay on or near the top all year of Web Analytics Blogs, regardless of how they're being measured; in  Top Ten Web Analytics Blogs: July 2007 Avinash Kaushik has Webmetricsguru as the #3 top Web Analytics Blog using a…

13. Judah Says:
    July 27th, 2007 at 15:16

    Thanks for the nice mention Avinash… :) Your blog is also one of my personal fav's as well!

    Namaste,

    Judah

14. Dennis R. Mortensen Says:
    July 27th, 2007 at 16:17

    Hi Avinash,

    Thank you once again for adding me to your list. This being a not-for-profit (actually with the number of Red-Bull’s consumed to write my blog, it is a for-cost activity) – it is always nice to get some recognition. Keep up the good work!
    
    A comment to Eric Peterson’s comment:

    - Hi Eric :-)

    When doing regression analysis (and any data analysis in general for that matter) – it is somehow understood that the more observations one provide the better the result (Avinash added, to my knowledge, a large pool of observations beyond the top 10 list itself). At the same time, if using only the FeedBurner dataset, and thus deploying a simple linear regression “analysis” as you suggest – it is again somehow understood that the more variables (that are associated) one can provide the better the result (Avinash added another variable and thus an opportunity to do a multiple regression analysis.

    So whether you love Avinash, me or Technorati or think any of them are flawed – does not change much about the fact that from a pure “scientific” point of view – Avinash took the better choice (compared to the old list and also compared to using Feedburner only) and that the list certainly improved.

    Cheers
    (…And I think I will see you in both Stockholm and Brussels soon, which I am very much looking forward to)

    Dennis R. Mortensen, COO at IndexTools
    My Web Analytics Blog

15. Dennis R. Mortensen Says:
    July 27th, 2007 at 16:49

    Hi Avinash,

    May I possibly suggest an update for your next list? – that you add a variable derived from the Google Ranking for the search term “Web Analytics Blog” – Assuming there is a decent association, you will be able to provide a list with higher “accuracy” and less dependency on the two variables you have now:

    - Influence (Technorati)
    - Content Value (FeedBurner)

    Google SERP for “Web Analytics Blog”
    Matt Belkin's blog | Inside Web Analytics
    Web Analytics Demystified
    Web Analytics Blog
    Coffee, Sun & Technology
    Official Google Blog: The circle of analytics
    VisualRevenue | Web Analytics & Affiliate Marketing blog
    Web Analytics .be Blog
    Occam’s Razor by Avinash Kaushik
    Site Visitor Tracking – Web Analytics Blog
    Mymotech is Michael Helbling's web analytics blog

    You could call this NEW variable:

    Content Relevance (Google)

    As this new varibable “punishes” e.g. Marshall – with posts beyond analytics and rewards the somewhat focused posts by Eric. Hence Eric’s 2nd position on the list and Marshall’s 48th position on the list. No offence intended Marshall! (just trying to add a valuable input to the list)

    This could principally be valuable on another point as well, namely that this is one list where Occam’s Razor is not at the very top of the list (assuming that your blog is one of those with the highest Technorati and FeedBurner numbers) – and this might even make Eric happy at the same time as well :-)

    Dennis R. Mortensen, COO at IndexTools
    My Web Analytics Blog

16. Eric T. Peterson Says:
    July 27th, 2007 at 19:36

    Hi Dennis,

    Yes, imagine my glee when Avinash had used linear regression in his new formula, something I think is a pretty good idea. It certainly seems to mitigate at least some of the problems seen in the Technorati-only strategy. But I still think Technorati is far more inaccurate than the 10% suggested in one of the comments above.

    Maybe it's like cookie-based visitor data?

    I like your "Content Relevance" suggestion as well — more data is better, plus it lets Avinash use a Google product (oh, except Feedburner is a Google product, never mind ;-) I don't, however, think Avinash should (or would!) change his methodology because it would make me happy (or Marshall unhappy) … that would be silly!

    (But thanks for thinking of me!)

    See you in Stockholm and Brussels!

    Eric Peterson
    http://www.webanalyticsdemystified.com

17. Steve Says:
    July 27th, 2007 at 22:03

    I've been chatting with Eric about this within his .. rebuttal posting, I suppose. :-) It seems only fair to add a comment or three to the blog of the author of this latest controversy. :-P

    To a large extent, I feel the methods used by yourself Avinash to generate this listing are somewhat secondary. I can make own mind up over which blogs I regularly read and which I unsubscribe from. That doesn't mean those I unsub are of less value, just of less value *for me*.

    Rather, where I feel this Top 10 WA Blogs series offers *real value*, is the explosion of quality blogs in the WA field. And a level of friendly rivalry to help push everyone to do better.
    eg Robbin's efforts to get everyone to shift their feeds and links across. She could have ignored that, but went the extra mile to get everyone to do so. A positive move all round!

    This push for excellence. That is the key value we see here.

    My 2c of course. :-)

    Cheers!

18. Michael Feiner Says:
    July 28th, 2007 at 03:55

    Hi Avinash,

    Certainly an interesting debate.

    Here are a few comments by a relatively new comer to this area. I have only recently developed a strong addiction to blog reading (going hand in hand with a personal seismic change from employee to entrepreneur).

    Look at how our industry is growing – fantastic to have so many WA blogs. I suspect by next year Avinash would have to publish a Top 50 ranking along side the Top 10.

    If the purpose of the ranking is to push everyone to do even better, as Steve suggests, then mission accomplished – great for us readers-only.

    The ensuing debate suggests a market opportunity. Aren’t the likes of BuzzMetrics, Cymfony and 1st2c in prime position to set up some standard measurement?

    I agree with Eric that accuracy is important.
    It seems to me, that reliance on feed subscriptions only wouldn’t necessary give us a true reflection of a blog’s “value” to the reader (I purposefully avoid using the terms “influence” or “engagement” as they will surely land me in trouble).

    Sure, TV ratings are predominantly about share of audience – who had more viewers. Reliance on one metric. Is that enough?

    Isn’t the recent debate over online measurement all about engagement rather than just visitor numbers. Case in hand – NetRatings adding total minutes per site to their measurements.

    Some companies are going to win some might lose. But are we to think that Google is a worse off advertising solution just because on average it gets less total minutes? I don’t think so.

    That is why I like Avinash’s idea of using a regression model, relying on more than one variable. I also think that Dennis’ suggestion of Content Relevance should be considered. It makes sense to me.

    Avinash, can I suggest adding yet another parameter to your regression formula. How about including the number of comments left by readers? This is a good measurement of interest and engagement with a blog. I’m not sure how easy it would be to collect this data (could be self-reported, same as the feed subscriptions).

    I concede that comments may be influenced by short term behaviour as you mentioned in your response to Bhupendra. It will also depend on how active the blogger is responding to readers’ comments and effectively inflating the number of comments.

    Actually is that a bad thing? If bloggers start responding even more often to comments than we (the readers) get more of a debate going – surely a good thing.

    Am I missing something really obvious? I’m sure you considered this option before.

    I intend to post a similar comment on Eric’s blog. So if you do decide to add the comments parameter into your calculation, well, at least my comments would be evenly spread.

    Thank you,
    Michael

19. Lars Says:
    July 28th, 2007 at 06:15

    Avinash,

    Despite the issues I have with Technorati there's no doubt your blog is #1 no matter how we look at it. :)

    I don't know that I agree with this statement though: "If a blogger chooses to post non-value added content to their blogs then no matter how great they are they will have little “influence”.

    I have come across many blogs (not related to web analytics) that have a ton of subscribers and high ranking but no unique content.

    Neither Technorati nor Feedburner necessarily say anything about quality. It's about being in the right place at the right time and about being smart about linking strategy etc. Look at MySpace — you can't tell me it's the best, yet it's huge.

    I agree that Dennis should make it to the top ten.

    I would like to see your personal top ten favorites as well as the quantitative top ten next time around.

    That would add a qualitative aspect to it, even if it's merely based on your opinion.

    =)

    Lars

    PS. It appears to be a good strategy to write about Google Analytics though there are some clear exceptions to that rule, I don't think Ian Thomas is writing that much about Google Analytics. ;-)

20. Judah Says:
    July 28th, 2007 at 08:12

    Another variable you could add to the regression could be the comment:post ratio expressed as a floating point integer. It hints at engagement with the blog.

    Judah

21. Avinash Kaushik Says:
    July 28th, 2007 at 11:07

    Dennis : Google no good. Wait that did not sound right (they are my client!).

    Google does not work. Oh that does not sound any better! My contract is being ripped right now. :)

    In this case Google does not work as awesomely as it does in all other cases. Ahhhhh much better. ^)

    Seriously though this is a great suggestion (and I think Marshall had suggested this to be a while back), but the problem is that some element of "freshness" is missing.

    Both the #1 and #4 blogs in that list have not published anything for the last year (and since both Matt Belkin and Xavier Casanova are thought leaders I think that is our loss not to that their contributions).

    I'll keep a eye out on this and see Google's algorithm improves for our specific application.

    Michael, Judah : Using comments per post is a lovely idea. I am a huge fan of "conversation rate" (more at: How to measure success of your blog.)

    The challenge I ran into was that this is easy for me to do on my blog's wordpress platform but on pretty much all other platforms (blogger, typepad etc) it is really hard for the blog owner to get this number, without actually manually counting it one by one.

    You are right about comment "inflation" by blog owner but it is a minor worry. If there is a easy way for blog owners to share their average comments per post in a given time period (say last three months) then I would take it in a giffy.

    I'll ping the blog owners for next round on this one (everyone was exceedingly kind in sharing their feed subs this time, so perhaps this will work after all).

    Thanks everyone.

    -Avinash.

22. Mambo number 5 « Web Analytics .be Blog Says:
    July 30th, 2007 at 04:37

    [...] Mambo number 5  As I was finishing up my meetings with René in Amsterdam on Friday afternoon – we are finally back in Brussels, after 2 weeks of France & The Netherlands – I read a mail coming from Mehdi on my BlackBerry, proudly announcing Avinash Kaushik had announced his Top Ten Web Analytics Blogs for July 2007 and our little pan European effort had ranked number 5. Lou Bega started singing in my head. [...]

23. Marianina Chaplin Says:
    July 30th, 2007 at 06:23

    Joining this debate a little late. Interesting to see this kind of debate but I would have to agree with a few of the skeptics that technorati is more about “reach” rather than “true influence”. I would be interested in an influence metric which would assess the number of comments on a site (even if these are inflated by the blog owner replying back to posts or trackbacks). But we may be quite a far way off from actually getting these kind of metrics as you said until other blogging platforms like blogger allow easy viewing of total comments etc like wordpress).

    Further down the road, we could calculate the true influence of comment flow (MIT are doing interesting work in this area) to take away the spammer adder-ons/trackbacks/own replys. Also number of characters in comments total – ie those encouraging real debate end up with lengthy comments on their blogs. And even further on, rating commenters with a visitor generated rating system (thumbs up/thumbs down – for starters). At the end of the day though, I think Avinash's blog would still be number 1 because he has this unerring ability to make the complex simple :)

    Cheers Marianina
    http://www.marianina.com

24. Nate Sidmore Says:
    July 30th, 2007 at 08:01

    Avinash,

    Didn't get a chance to check back into the conversation over the weekend as I'd hoped, but one late observation is that based on the holes in Technorati described by Eric I suspect the margin of error might be higher than 10%, which he mentions above. Yet I do agree with the post someone mentions on his blog that there is no other bar being set to measure your list's quality against, so this is the best we can get (and it's really good stuff at that).

    But you've been given plenty of suggestions for improvement etc, so I'll let the overall conversation lie (lay?). It does bring up the related question in my mind: Is there a point at which data becomes too imprecise to be any use at all?

    If you have a chance to react to that question, great, but not a huge deal if not.

    Thanks for the forum,

    Nathanael

25. Alex Says:
    July 30th, 2007 at 09:51

    I find it interesting that no one has choosen to focus on the fact that regardless of which of the additional suggestions are used the ranking will stay pretty much as it is above (it shifts just a bit).

    Try it. As an example, I just finished several hours counting comments on each of the blogs listed on the top ten.

    Thanks Avinash for keeping the conversation respectful and for putting up all the effort that surely goes into producing something like this (not to mention each post you publish).

    Alex

26. Ralf Haberich Says:
    July 31st, 2007 at 00:46

    I wonder, when a German blog will be mentioned here…

    Great Work, Guys.

    ralf haberich.
    http://www.webanalyticsblog.de

27. Top Blogs in Web Analytics – Mymotech Says:
    July 31st, 2007 at 09:41

    [...] Avinash Kaushik has recently published his top 10 Web Analytics Blogs. And while I am not on that list, it is interesting to see the emerging debate that it has sparked. The ranking factors for Avinash's list are comprised of some wrangling of Feedburner subscribers and Technorati rank. But let's face it ranking anything in the world of analytics is like throwing red meat to an audience of hungry rottweilers. So kudos to Avinash for a brilliant idea. While I do not expect to be on Avinash's list any time soon, it is nice to see that the number and quality of web analytics blogs is gowing and changing. This is such a great industry to be a part of. [...]

28. Kevin Hillstrom Says:
    July 31st, 2007 at 16:21

    It has been very interesting to read all twenty-seven comments submitted, to date. I doubt I've had 27 comments on my blog during the past 27 days!!

    I have always struggled with how to create a ranking system that is "fair", one that represents reality.

    Any equation I've developed fails to represent reality. Technorati tells us what other bloggers think of your content. Alexa tells us what a small sample of folks who use the Alexa toolbar visit. Feedburner stats are intriguing, because it counts subscribers — presumably, the most loyal followers of any blog.

    For me, Google represents "customer acquisition". If Google likes a site, it sends pre-qualified traffic to the site. These visitors might become repeat visitors, e-mail subscribers, or RSS subscribers.

    So, there's lots of important elements that go into a ranking system.

    And yet, when you plug in all the "right" pieces of information, you often end up with a ranking system that is very similar to one with a small number of "predictors".

    Ultimately, you try to develop something that is simple, yet captures "what is important". Hopefully, what Avinash is advocating will provide visibility for the blogs he's tracking. And at the end of the day, that's what is really important. It is important to make sure that the folks who provide valuable content get noticed!

29. what is interesting this week (July 31th 2007) at Blogjer is about blogger Says:
    July 31st, 2007 at 18:46

    [...] Eager to know what is the best web analytic blog? Avinash has the answer [...]

30. Die Top-10 Analytics Blogs | eControlling.de Says:
    August 7th, 2007 at 03:34

    [...] Der Analytics-Blogger Avinash hat ein Ranking der Top-Web Analytics Blogs für den Juli 2007 veröffentlicht.
    Die Top-10 Blogs werden algorithmisch ermittelt, dabei reicht ein hohes Ranking bei Technorati jedoch nicht aus- ein wichtiges Kriterium beim Ranking sind die zahlreichen Feed-Subscriber. Homepage: Link [...]

31. Lars Says:
    August 10th, 2007 at 13:54

    I don't think Google rankings are a good way of measuring how good a blog is (like Dennis suggested). Why? Well, I could have a crappy blog and master black hat SEO and rank as #1.

    The Excel file only asks for Feedburner (or the equivalent) and Technorati data, where does Alexa enter the equation?

    The black hat SEO blog would get many first time visitors but probably not that many return visits which hopefully would show in Alexa rankings.

32. Dr. Mohamed Taher Says:
    August 15th, 2007 at 21:12

    You have a real creative mind. I envy your a) holistic approach to the area of Web analytics, and b) ability to express your passion in a way that really communicates, and communicates well. You get ten out of ten for this information visualization. Keep up the good work.

    Best wishes from a Information Visualization blogger and a Multifaith Web analytics' beginner.

33. Kango Blog » Blog Archive » Blog Day 2007! Says:
    September 1st, 2007 at 00:59

    [...] 3. Occam's Razor – I've watched from afar as former colleague Avinash Kaushik was the Peter Parker of Intuit's Web Commerce Group but then became the Spiderman of Web Analytics.  Here's an example of both great blogging and great analytics, two great tastes in one.  This post cured me of my desired to get Dugg.  Quality over quantity. [...]

34. Rick Says:
    December 7th, 2009 at 00:05

    Very good!The that you’re actively working to improve the quality of your ranking — that shows real commitment on your part and you’re to be commended for that. And I do get that, as one commenter.

Join the conversation. . . .