By Ellie Fields 2008/12/09

Is a top-viewed landing page catapulting visitors away from your site? Is a lesser-viewed page making your visitors insanely happy? Use Google Analytics data to find out.

If you’ve used the Google Analytics Top Content Report, you know there’s a lot of data there. You also know how frustrating it is to comb it all. This week I was trolling Tableau’s web analytics data to find the fast and the lame.

Lots of Data, But Ugly

Google Analytics’ content report is generous with data: pageviews, time on page, bounce rate, and % of visitors who exit the site after viewing the page. But the report doesn’t exude insight. In fact, it can make your head hurt:

Web Analytics Can Be Beautiful

So try it graphically. Here’s our website’s content, organized by section (the first level of the URL hierarchy).

The visualization tells me a few things:

  • The products section gets the most pageviews. Because Google Analytics breaks out each landing page individually, this type of aggregate data is hard to get.
  • The high-pageview sections—the homepage (/), learning, community-- all have green bars, which means visitors tend to stay on the site after viewing them. From this we can guess they found content they liked.
  • Visitors spend more time in videos than anywhere else.
  • But then the video viewers leave the site. The video bar is a dark, angry red, showing a high exit rate.

So our video-watchers show commitment by spending a lot of time on the page, but then leave the site. Our sales and support guys often send customers to particular training videos, so it’s natural that they might leave after getting what they needed. Still, we could do a better job of providing relevant next-step content on those pages.

How Did Our Landing Pages Do?

Now we know how visitors are interacting with different parts of the site. Let’s check out specific content. I drilled into Tableau’s gallery to evaluate our Visual Examples, which are visualizations from customers or from public data.

What popped in this visualization:

  • The visual analysis of baby names got the most pageviews, and had a low exit rate(it’s dark green). So people liked that example well enough to keep going, probably into more examples.
  • The analysis of auto sales and the sales pipeline visualization had the next highest pageviews, but high exit rates. It’d be worth asking some customers if they got what they expected from those examples.
  • People spent the most time reading about hurricane tracks. I checked out the visualization and it was immediately clear why-- turns out we had forgotten to include one of the color keys. People spent time on the page because they were trying to figure it out. We’ve fixing the problems.

To make Google’s content report visual, I simply exported the CSV, formatted it slightly in Excel and then connected it to Tableau (don’t have Tableau? Download a free trial.) I didn’t use the Google Analytics API, but I’ll look into that in the future.

If your content report is filled with unreadable URLs, the Google Analytics Blog offers tips on cleaning them up. And if you use query strings to drive content, Analytics Talk shows you how to make them more readable.


Hi Ellie,

Now you got my attention! I'm a Web Analytics expert, and Tableau lover, although I have never had the opportunity to use it for specific WA work.

I would be very interested in learning how Tableau does with GA API (you have access?). Also, have you tried with WebTrends or Omniture data?

I find what you're finding with Tableau compared with GA is quite interesting. Have you tried any other reports in GA?

The Google Analytics pages are pretty cluttered, but my "important" data I download for offline analysis anyway. I use Excel, but I'm intrigued by Tableau's use of color to highlight the % leaving the site from the given page or section.

I'll bet the baby names got a lot of hits through the search engines by people expecting a child, not looking for infographics. When they got to your site, they found other interesting topics to hold their attention. Graphics, especially done well, attracts viewers.

Hello Jacques!

I'm planning to use GA for a lot of web analytics, though content is the first section that I dove into in depth. I have not yet used Webtrends or Omniture data, though that would be interesting.

GA reports are very hard to consume, and we have that data internally, so I started there. I'd like to move onto the API but have not yet- this was downloaded to Excel. Have you used it successfully?

@Jon: Yes, I think you're right about the web searches for baby names. Every new parent I know did some web searching when they were thinking of names.



I've used Tableau with Omniture data. It's fast, and signifcantly more so than Omniture's own SiteCatalyst software. I competed in a web analytics competition and Tableau shaved around 20 hours off the average analysis time, as compared to the other contestants. Though Omniture's data is exhaustive, the visual analysis and ad hoc capabilities as well as the ability to compare two nominal variables are really poor. Tableau enabled me to do that in seconds. I took the Omniture data, exported it to a huge Excel file I created, and then performed the analysis.

Hi Ryan,

Very interesting! Did you have to export specific reports from Omniture, and then concatenate them, or could you export the entire report? How did you find working with Omniture's aggregation of data, which I guess would limit what dimensions you could compare and how.

Hi Jacques,

I did have to export the specific reports (instead of one comprehensive report) I wanted from SiteCatalyst to a really large Excel spreadsheet and then connect that to Tableau. I'm not sure if Tableau could be directly hooked up to Omniture site data, because Omniture may have some restrictions. Due to the nature of the competition I was unable to do that at least.

Obviously, if you have worked with SiteCatalyst before you know that there are limitations on the reports you can pull. However, pulling reports, aggregating into Excel and then analyzing using Tableau enabled me to look at comparisons and trends across categories that Omniture doesn't allow (at least not without doing extra work to create customized reports).

Hopefully, that addresses your questions. If not, please clarify and I will get back to you.

Does anyone know if Tableau hooks into ODBC connections, specifically the ODBC drivers that come with WebTrends??

I have had a look but could not really see anything that says they can?

We have an Excel spreadsheet that does it, so I guess I could use that, but ideally I would like to cut out the middle man and just use Tableau to connect direct to Web Trends.

i want to impliment Tableau Software in my syte. but i am not aware about it,
can you please let me know how to implement it in a web site using along with, weather we have to use API for it , if yes from where we can find it.
thank you
dinesh chandra

Hi Luke,
Tableau does not support generic ODBC connections, however you can use Access to set up a linked table that uses the ODBC connector for your database. Tableau can then connect to the Access database, allowing you to sidestep the process of creating an Excel file each time. Note however that this approach can have performance and functionality problems, and we recognize that it isn't the ideal solution. I believe this has been discussed in the forums, so if you have problems I suggest you read others' recommendations and post any questions you have there.