{{ searchResult.published_at | date:'d MMMM yyyy' }}

Loading ...
Loading ...

Enter a search term such as “mobile analytics” or browse our content using the filters above.

No_results

That’s not only a poor Scrabble score but we also couldn’t find any results matching “”.
Check your spelling or try broadening your search.

Logo_distressed

Sorry about this, there is a problem with our search at the moment.
Please try again later.

A fantastic post by Francois Derbaix, CTO at top French travel site Toprural.com, tells us that Google is as susceptible as anyone else to imposing conditions that make its services look good when you analyse your site traffic.

Comparing the stats for Toprural delivered by Google Analytics with his own 3rd-party solution, he finds that, while his own system (AT Internet’s XiTi) says 37.8% of visitors come via Google, GA says it’s 71.8%. The core of the problem, he discovers, is the good old cookie window. It turns out the default cookie window Google ascribes to visitors that arrive on a site via Google is six months. Six months!

So, every visit a user happens to make to a site for the six months after having once visited that site via Google is ascribed to the search engine. The standard window for display media is 30 days, a sixth of the length Google has opted for, while we’re typically asked to assign search cookie windows of a couple of days at most.

Monsieur Derbaix gives an example of how the cookie window might impact your site stats (the extra joy of this quote is a demonstration of how Babelfish has translated his post from French to English): Example: Pierrot visits your site 2 times in a direct way (it types your URL in the bar of address of the navigator), after which it comes there once through the results from a research on Google. Then he visits the site 3 times still on line, without passing by again by Google, in the 6 months which follow.

“How much visits since Google, GA will indicate? 4 (1 + the 3 following ones). How much direct visits indicated? 2 (two first). Whereas actually there is 1 well visit by the means of Google and 5 natural/direct visits.”

In other words, in his example, a user that actually visited the site via Google just once, is credited to the search engine four times.

It’s crucial to point out that the six-month window is just default and configurable inside GA, but it’s safe to assume that few GA users will a. be aware/understand the importance of this setting or b. be able/willing to do anything about it.

For me, this is an indicator of lots of things: first, perhaps more explanation of why Google launched Google Analytics for free in the first place; second, the real importance for site owners to understand the detail of how online advertising works; in a detail like cookie windows, for example, lies crucial information; and, third, the importance of independent providers in helping agencies and advertisers make sense of the wealth of info now at their disposal.

If you make decisions based on data, then that data has to be reliable.

Paul Cook

Published 5 June, 2009 by Paul Cook

Paul Cook, the founder of RedEye and TagMan, is a contributor to Econsultancy.  

28 more posts from this author

Comments (16)

Comment
No-profile-pic
Save or Cancel
Avatar-blank-50x50

nivla

Would be great if a GA representative could offer his/her views on this. Any chance of posting their response next post? Would def. offer a useful balanced take.

about 7 years ago

Adam Tudor

Adam Tudor, Senior Digital Marketing Manager at The Black Hole

"check the quality of your own website metrics and customer data."

Totally agree - implementation of analytics can give you a good overview as to site activity, but making sure all aspects are fully understood is critical if it is being used in part to make business decisions.

All web analytics packages aggregate visitors, sessionisation, differently, have different cookie and referral lengths, and do things slightly different - it's critical to understand what constitutes a visitor in the package you are using if you are doing detailed analysis.

about 7 years ago

Rob Jackson

Rob Jackson, UK Managing Director at Elisa Interactive Ltd

A lot of people assume because GA is free and made by Google it is an off the shelf solution for their site. The fact is a lot of the best GA features have to be set up or customised to a client's specific needs. This is why Google has setup the GAAC network so that there is a support network of highly skilled analytics pros to avoid issues like this.

about 7 years ago

Avatar-blank-50x50

Andrea Giannangelo

Hi,

could anybody give a clear explanation about how to configure GA to avoid this problem?

This is a really important problem, I will be thankful with the one who will answer.

Greetings,

Andrea Giannangelo

about 7 years ago

Avatar-blank-50x50

Russell Smith

Great post! Thanks for bringing this to the forefront. I knew the default cookie window in GA was six months, but didn't realize the full implications.

about 7 years ago

Avatar-blank-50x50

Jiri Brazda, Founder at Optimics (www.optimics.cz)

Hi, my understanding is that the same 6-month cookie is set for whatever campaign (traffic source), not only Google itself. So yes, the GA reports have to be read with this fact in mind. In the example above, if one of the three consecutive visits to the site were from Yahoo (generally, anything but direct access/bookmark) they would be attributed to Yahoo, and not Google.

I can't read French but I think the post is a little misleading and creates panic which is uncalled for. It it was true, however, it would be a dead serious reason to stop using GA immediately.

However, this post opens up a debate regarding traffic/conversion attribution in which area I agree GA is badly lacking since it prefers last-click attribution and distorts the fact that a purchase (or any conversion in general) often is not a 1-click affair but consists of a number of visits driven by a number of campaigns/traffic sources.

Let's keep the debate going!

Jiri Brazda

about 7 years ago

Avatar-blank-50x50

Adrian

Paul, Catchy headlines are good for traffic by your selective highlighting of facts is disappointing. In the post that you reference Avinash Kaushik had posted a detailed comment outlining exactly how Google Analytics works and that the supposed "bias" has nothing to do with Search. Here is his comment on Francois Derbaix's blog: http://francoisderbaix.com/2009/02/17/google-analytics-vs-xiti-como-miden-el-origen-de-las-visitas/#comment-667 Here is a great way to think of how Google Analytics: The last campaign gets “credit”. For the purposes of the discussion here campaigns include Email, Search, Affiliates, Display, Video, Social Media, Whatever else catches your fancy. Here are some scenarios of how it works (across multiple visits): Direct, Email, Convert. Credit: Email. Email, Direct, Convert. Credit: Email. Email, Paid Search, Organic, Convert. Credit: Organic. Affiliate, Paid Search, Email, Convert. Credit: Email. Google Paid Search, Microsoft Live Organic, Credit. Credit Microsoft Organic. Google Paid Search, Email Campaign, Affiliate, Convert. Credit: G Paid Search=1 visit, Email=1 visit, Affiliate=1 visit. Conversion=Affiliate. Email, direct visit, direct visit, direct visit. Credit: Email – 4 Visits. Yahoo!, direct visit, direct visit. Credit: Yahoo – 3 visits. Direct. Credit: Direct. I hope this scenario outline helps explain how things work in Analytics. You can swap any campaign above for any other you are doing.

about 7 years ago

Avatar-blank-50x50

Jason Green

Also important to point out is that if your visitor arrives as "Direct" traffic but later visits your site through a search engine, GA overwrites "Direct" with "<search engine of choice>" as the source of the traffic.

about 7 years ago

Avatar-blank-50x50

François Derbaix

@ Paul Cook: thank you very much for commenting about my analysis (posted on my blog in FR and ES)! I am glad the debate can continue now also in English on Econsultancy.com.

About this umtz cookie set by default to 6 months validity period, it is also interesting to know that experts can modify it and set it to a lower expiry period if they want. This is no easy to modify (Google could have made it more customizable) but they do publish the documentation about how to change the GA scripts.


Anyway my personal conclusion is that GA Sources statistics is good for conversion analytics purposes but does not provide you with a real statistic about your sources and the % on which your traffic depends on Google. Direct traffic is largely underestimated. I think it's important to keep it in mind when reading GA data.

about 7 years ago

Avatar-blank-50x50

Don

Appreciate this post as I have just kick started the process of testing other Analytical packages on some of my sites. As one gets deeper into this theme on realises that there is so much more to learn.

about 7 years ago

James Sandoval

James Sandoval, SVP & Managing Director EMEA at Signal

Nice posts Paul/Francois. Got me thinking. No time to write a proper comment other than this. Will try later. James

about 7 years ago

Avatar-blank-50x50

Adam Marshall

He's dead right on the mark by saying:

"It’s crucial to point out that the six-month window is just default and configurable inside GA."

which is the true value of the GA system IMHO, especially coupled with the ability to set multiple profiles for the same site (and/or segment thereof) with different configuration settings.

Although a lot of people seem to enjoy 'dissing' GA, and they often have valid points, it's this kind of in-built flexibility that still makes me very fond of the platform.
(other key factors being the in-built free-ness, and the usability of the interface)

about 7 years ago

Avatar-blank-50x50

Charles Burrows

Interesting post, though it doesn't seem as sinister to have the 6-month cookie if you accept it uses that window to attribute all acquisition channels.

We are currently building GA into our analytics and I'm impressed with it as a free (at point of implementation) service.

about 7 years ago

Avatar-blank-50x50

duncan black

Google also make it difficult to find out exactly what people have typed into the search engine when clicking on your paid google adwords.  For example, if you're bidding on the keyword 'website design' analytics will show that you have had xx visitors on your site from your campaign who typed in 'website design' whereas when you setup a couple of filters you can see what people actually typed in i.e. 'free website design', 'website design courses' etc.

I think the idea behind this is that if people could actually see the irrelevant clicks they were paying for clearly and easily, they would go into their adwords accounts and add more negative keywords thus reducing googles advertising revenues considerably.

about 7 years ago

Matt Isaacs

Matt Isaacs, Founding Partner & CEO at EssenceEnterprise

@Tom Simpson - excellent summary of the issues.  Particularly with respect to overreliance on adservers - which are after all really built for serving ads, not tracking.  Having said that I have seen adserver implementations where they are genuinely configured to provide holistic tracking across all digital activity in a transparent and robust way - but that appears to be the exception not the norm.  Whether by accident or design adserver tracking is all too often configured to make paid for marketing, and particularly display, seem better than it really is.

@ Duncan Black - you can obviously also just run a search query report in Google Adwords which would give you the detail on actual search terms used (and is what most PPC search marketers use in my experience)

about 7 years ago

Paul Cook

Paul Cook, Director at NCC Web Performance

Nail on the head there @Matt Isaacs and @Tom Simpson. The reliance on ad servers (or ad network or affiliate network or search provider) for tracking is a real issue. Client-side solutions are now essential to extract and make sense of all the data these systems provide. Ahem.

about 7 years ago

Comment
No-profile-pic
Save or Cancel
Daily_pulse_signup_wide

Enjoying this article?

Get more just like this, delivered to your inbox.

Keep up to date with the latest analysis, inspiration and learning from the Econsultancy blog with our free Daily Pulse newsletter. Each weekday, you ll receive a hand-picked digest of the latest and greatest articles, as well as snippets of new market data, best practice guides and trends research.