Comments on: Estimating the effects of cookie-deletion http://showmeanalytics.com/2009/04/calculating-the-effects-of-cookie-deletion/ Analytics from the Show Me State Mon, 05 Jul 2010 14:33:08 -0500 http://wordpress.org/?v=2.8 hourly 1 By: Fréquence : prendre en compte la suppression des cookies | Blogue.PubInteractive.ca http://showmeanalytics.com/2009/04/calculating-the-effects-of-cookie-deletion/comment-page-1/#comment-671 Fréquence : prendre en compte la suppression des cookies | Blogue.PubInteractive.ca Mon, 03 Aug 2009 18:59:09 +0000 http://showmeanalytics.com/?p=75#comment-671 [...] sera bien moindre sur un site sur lequel les internautes ne vont qu’une seule fois par mois. Voici ici un article intéressant sur la question (EN). Je pense qu’il est préférable d’ajuster un peu à la baisse le cap de [...] [...] sera bien moindre sur un site sur lequel les internautes ne vont qu’une seule fois par mois. Voici ici un article intéressant sur la question (EN). Je pense qu’il est préférable d’ajuster un peu à la baisse le cap de [...]

]]> By: Trending Upward | More Great Analytics Resources http://showmeanalytics.com/2009/04/calculating-the-effects-of-cookie-deletion/comment-page-1/#comment-171 Trending Upward | More Great Analytics Resources Thu, 28 May 2009 01:32:52 +0000 http://showmeanalytics.com/?p=75#comment-171 [...] up is showmeanalytics.com. You might know my opinion toward unique visitors and Angie’s last couple posts have been around this topic. Very interesting. I love the debate over unique [...] [...] up is showmeanalytics.com. You might know my opinion toward unique visitors and Angie’s last couple posts have been around this topic. Very interesting. I love the debate over unique [...]

]]> By: David http://showmeanalytics.com/2009/04/calculating-the-effects-of-cookie-deletion/comment-page-1/#comment-160 David Wed, 20 May 2009 18:54:57 +0000 http://showmeanalytics.com/?p=75#comment-160 comScore has been grappling with this for some time now, and you can check out the cookie deletion white paper here: http://www.comscore.com/Press_Events/Presentations_Whitepapers/2007/Cookie_Deletion_Whitepaper comScore has been grappling with this for some time now, and you can check out the cookie deletion white paper here: http://www.comscore.com/Press_Events/Presentations_Whitepapers/2007/Cookie_Deletion_Whitepaper

]]> By: Jim Williams http://showmeanalytics.com/2009/04/calculating-the-effects-of-cookie-deletion/comment-page-1/#comment-159 Jim Williams Wed, 20 May 2009 17:49:45 +0000 http://showmeanalytics.com/?p=75#comment-159 Angie - great post. You have prompted me to calculate the visitor inflation for our site - a job I have been avoiding for sometime. Not quite sure where I am going to get the deletion numbers from upon which to base my assumptions though. I saw a comScore paper sometime ago - trouble is our visitors are generally teenagers and we have strong evidence that they are visiting on multiple machines at home and at school. We also make the assumption that cookies are deleted from school machines after every session - although we have no evidence to back this up. Luckily we do have registration data which would indicate that the top end for visitor inflation must be somewhere in the region of 50% but we must do a more formal analysis. Angie – great post. You have prompted me to calculate the visitor inflation for our site – a job I have been avoiding for sometime. Not quite sure where I am going to get the deletion numbers from upon which to base my assumptions though. I saw a comScore paper sometime ago – trouble is our visitors are generally teenagers and we have strong evidence that they are visiting on multiple machines at home and at school. We also make the assumption that cookies are deleted from school machines after every session – although we have no evidence to back this up. Luckily we do have registration data which would indicate that the top end for visitor inflation must be somewhere in the region of 50% but we must do a more formal analysis.

]]> By: Dan http://showmeanalytics.com/2009/04/calculating-the-effects-of-cookie-deletion/comment-page-1/#comment-78 Dan Mon, 20 Apr 2009 17:06:58 +0000 http://showmeanalytics.com/?p=75#comment-78 Wouldn't we be more interested in the trend anyway, rather than the actual number? Wouldn’t we be more interested in the trend anyway, rather than the actual number?

]]> By: angie http://showmeanalytics.com/2009/04/calculating-the-effects-of-cookie-deletion/comment-page-1/#comment-76 angie Mon, 20 Apr 2009 12:03:38 +0000 http://showmeanalytics.com/?p=75#comment-76 Jake, I agree, and it's another reason I can't get behind requiring a cookie-deletion algorithm for web analysts or for the tools they use. What we use now is bad, no doubt, but a "correction" won't necessarily make it better. There are too many facets to this problem: cookie deletion, multiple devices, and multiple browsers pulling numbers one way, while shared logins and shared computers (big issues for several of my B2C sites) pull the numbers the other way. The only ways I can think of to get better unique visitor numbers come with a hefty price on privacy, real or perceived. There are definitely no easy answers. Jake, I agree, and it’s another reason I can’t get behind requiring a cookie-deletion algorithm for web analysts or for the tools they use. What we use now is bad, no doubt, but a “correction” won’t necessarily make it better. There are too many facets to this problem: cookie deletion, multiple devices, and multiple browsers pulling numbers one way, while shared logins and shared computers (big issues for several of my B2C sites) pull the numbers the other way.

The only ways I can think of to get better unique visitor numbers come with a hefty price on privacy, real or perceived. There are definitely no easy answers.

]]> By: jake http://showmeanalytics.com/2009/04/calculating-the-effects-of-cookie-deletion/comment-page-1/#comment-73 jake Mon, 20 Apr 2009 03:43:37 +0000 http://showmeanalytics.com/?p=75#comment-73 Cookie deletion is indeed a factor, but a similar analysis could be done with multiple devices too. Mobile devices are making this especially apparent. I might be a bit of an outlier, but just today I've hit my normal round of sites on my work laptop, iPod Touch, home laptop, and home desktop. 1 person, 4 cookies, and no way to tell of any overlap, unless authenticated. Throw in multiple users on one device (my home desktop and laptop are shared) and you've got quite a mess that I just can't see any algorithm overcoming. It would be as arbitrary as UV's already is. Cookie deletion is indeed a factor, but a similar analysis could be done with multiple devices too. Mobile devices are making this especially apparent. I might be a bit of an outlier, but just today I’ve hit my normal round of sites on my work laptop, iPod Touch, home laptop, and home desktop. 1 person, 4 cookies, and no way to tell of any overlap, unless authenticated. Throw in multiple users on one device (my home desktop and laptop are shared) and you’ve got quite a mess that I just can’t see any algorithm overcoming. It would be as arbitrary as UV’s already is.

]]> By: angie http://showmeanalytics.com/2009/04/calculating-the-effects-of-cookie-deletion/comment-page-1/#comment-72 angie Sun, 19 Apr 2009 22:59:25 +0000 http://showmeanalytics.com/?p=75#comment-72 Thank you Chris! Ned, that's a very important point: the visitor frequency (daily, weekly, monthly) from our analytics tools is completely out of whack due to cookie deletion. And it's why I think developing a real cookie-correction algorithm isn't going to be an easy task. I suspect that if someone is good with matrix algebra, they *might* find that decent visitor loyalty numbers could be estimated from WA tool numbers given assumed cookie-deletion rates, but I haven't taken it that far, and I no longer consider myself to be "good with matrix algebra." Thank you Chris!

Ned, that’s a very important point: the visitor frequency (daily, weekly, monthly) from our analytics tools is completely out of whack due to cookie deletion. And it’s why I think developing a real cookie-correction algorithm isn’t going to be an easy task.

I suspect that if someone is good with matrix algebra, they *might* find that decent visitor loyalty numbers could be estimated from WA tool numbers given assumed cookie-deletion rates, but I haven’t taken it that far, and I no longer consider myself to be “good with matrix algebra.”

]]> By: Ned Kumar http://showmeanalytics.com/2009/04/calculating-the-effects-of-cookie-deletion/comment-page-1/#comment-70 Ned Kumar Sun, 19 Apr 2009 17:42:02 +0000 http://showmeanalytics.com/?p=75#comment-70 Great post Angie and a good analytical excercise. The one thing I would caution folks is not to use the cookie output to determine the % of daily, weekly, and monthly visits -- I know it is common sense but sometimes we fall into the catch22 situations :-) Unfortunately, this is is an issue. I have found from my own excercise that the inflation can up up 200% in some cases. The good news was that I found the cookie inflation to have a variation among various segments of customers and so at least you know which one is relatively accurate and which one you have to take a with a bucket of salt (or do some mathematical wizardry like Eric had mentioned in one of his posts). In addition to the technique you mention above, an alternative way to get an estimate would be to use your "known" visitors (with a login id or something similar) to draw parallel conclusions for your site. With cookies, one thing I have found is that a triangulation method is the best approach to attack this problem -- coming at it from various angles and then sythensize the results for an action plan. Again, nice job with the explanation. Great post Angie and a good analytical excercise. The one thing I would caution folks is not to use the cookie output to determine the % of daily, weekly, and monthly visits — I know it is common sense but sometimes we fall into the catch22 situations :-)

Unfortunately, this is is an issue. I have found from my own excercise that the inflation can up up 200% in some cases. The good news was that I found the cookie inflation to have a variation among various segments of customers and so at least you know which one is relatively accurate and which one you have to take a with a bucket of salt (or do some mathematical wizardry like Eric had mentioned in one of his posts).

In addition to the technique you mention above, an alternative way to get an estimate would be to use your “known” visitors (with a login id or something similar) to draw parallel conclusions for your site. With cookies, one thing I have found is that a triangulation method is the best approach to attack this problem — coming at it from various angles and then sythensize the results for an action plan.

Again, nice job with the explanation.

]]> By: Christopher Berry http://showmeanalytics.com/2009/04/calculating-the-effects-of-cookie-deletion/comment-page-1/#comment-69 Christopher Berry Sun, 19 Apr 2009 17:12:05 +0000 http://showmeanalytics.com/?p=75#comment-69 Timely and relevant! A very serious attempt to estimate the size of the problem. Timely and relevant!

A very serious attempt to estimate the size of the problem.

]]>