Friday Jan 04, 2008

Report Envy - Making Omniture Look Like Google Analytics

I wish I could merge Google Analytics and Omniture together. Just put the polished UI and responsive search times of Google Analytics over the more powerful functionality of Omniture. Take for example the Top Content report for Google Analytics:

Classic Google design - has all the vital information in one screenfull by default. So how to get this in Omniture?

When you go to Paths > Pages > Pages > Most Popular Pages, you just get the Page Views. You can add more columns to the report by clicking Page Views in the header of the Details column and adding the following to the report:

Then for the Bounce Rate, you need to define it as a Calculate Metric. Click the Calculated tab and then click the Define New Metric button. Name it Bounce Rate, set the Metric Type to Percent, and enter [Single Access]/[Entries] in the Formula field.

Then add Bounce Rate to your report as well and viola, you've got a nice Google-esque Top Content Report.

A quick note on possible uses of this information. We've got two types of pages that we track the most - informational pages like videos and tutorials, and index pages that lead people to the information. So we look closely at the bounce rate for the pages. If a tutorial has a high bounce rate, that's not necessarily bad. The user may have found what they were looking for and then left the site. So we look at both the time spent and the bounce rate. A tutorial with a high bounce rate and a low time spent on page means the user probably did come to the page, say yuck, and leave completely.

When this happens it's time to look at the searches that brought people to the page. Google is again better off in this because you can see the bounce rate for each search term, so you can see which ones are making people go yuck the most. Not sure how to get this running in Omniture yet.

Of course, index pages with high bounce rates are big warning signs as well. When that happens it's time to look at the design of the page altogether.

Monday Nov 26, 2007

New vs. Returning in Omniture

One of the things that Google Analytics does really well is New vs. Returning users. I really like this metric, because it shows you a lot about how different users are using your site.

So this is where I admit my ignorance and ask a bunch of questions rather than give great insight. While New vs. Returning is cool, it really becomes powerful when you can use it as a segmentation tool. Google Analytics lets you do some segmenting for new vs. returning through the drop-down in the New vs. Returning report. The things you can segment are the most important high-level metrics - conversion rate, bounce rate, visits, etc. But it would be nice to allow you to segment more reports by New vs. Returning - things like most popular landing pages, most popular pages, etc.

Now Omniture is the self-proclaimed king of segmentation. Using Discover you can (supposedly) segment anything by anything (as long as the two reports you're using are looking at similar data). But unfortunately, we're having one hell of a time getting New vs. Returning even to show up, much less use it as a basis for segmentation.

Some of the red herrings we've been chasing:

  • Going into Discover and filtering by Visit = visit number does not equal one. Sounds good, but the problem is that if you run the report for the last six months, it only shows you new vs. returning for those six months. So there's no way to do "visit number does not equal one" outside of the range that you're actually generating reports for.
  • Just using Yearly Return Visitors. I'd be fine with just counting returning visitors in the last 12 months rather than "ever". After all, most people delete cookies more than once a year, and with our site, if you haven't been here in over 9 months, you might as well be a new user. Problem is that Omniture seems to count Yearly Unique Visitors as in this calendar year rather than in the last 12 months. Now that's just down-right dumb. Who in web metrics cares about calendar years? You care about trends! And your trends are all messed up because you'll show a huge number of new visitors in January and a low number in December.
  • Doing the math by hand taking all unique visitors and subtracting new visitors. Gets you the number but you can't use if for segmentation.

Now it looks like you can get the info in your conversion reports under Visitor Loyalty, but we haven't been using the Conversion reports because our main conversion metric - Downloads - hasn't been under Omniture tracking for technical reasons. We've finally figured that out so I'll post the results of that quest when I get it set up.

But the long and the short of it is - this really should be much easier in a top-of-the-line pay product like Omniture. If anyone has experience doing this or tips and tricks, feel free to let me know!

Friday Nov 16, 2007

Regular Expressions Anywhere in Google Analytics

So this is just a quick tip that may have been obvious to everyone but wasn't to me and I found it incredibly useful. In Google Analytics, you can use regular expressions in any search field. So if let's say you have an ecommerce site that dynamically forms URLs like this:


you can go to Top Content and enter a regular expression like this in the search box:


Then if you wanted to see all orders for more categories you just replace type=book with something like type=(book|cd|dvd).

Of course, if the site had been thinking about analysis from the beginning, they would have formatted that URL to put type=book before the id, so that you could just search for the most static parts of the URL (thanks.php?from=order;type=book) and not have to use a regex in the search.

So all you webmasters out there, do the analysts a favor! When you formulate your dynamic URLs, the parts that are the most dynamic should go at the end. So if your information looks like this:

      Individual ID

That's the order they should go in in the URL. Makes it much much easier to generate reports.

Another thing I'm figuring out is that I'm going to have to learn regular expressions better. People on the lists are pretty helpful but I think I need to just bite the bullet. Anybody got any advice on a good crash course (preferably online and free) to regex?

Thursday Nov 15, 2007

Round 3 of My Blog

Welcome to the third incarnation for my blog. It's been an interesting ride. My first round at blogging was when I was a technical writer here at Sun Microsystems and I focused mostly on tips and tricks for using NetBeans IDE. Then as our team started making more and more screencasts and demos I reinvented the blog as a Flash blog.

Now that I'm the manager of the NetBeans docs team, unfortunately I no longer have much time to play around with cool technology. But I do have new challenges, such as making sure the NetBeans web site is doing its job: informing users on how to get the most out of NetBeans, driving volume for Sun's technologies and tools, and fostering the growing community of NetBeans and Java users.

Which led me to a new semi-technical passion - web analytics. Digging through Google Analytics and Omniture reports, trying to figure out which key performance indicators really measure success, doing change-measure-change improvements on the web site. I have to say it's a little scary that I like this so much - it seems to appeal to parts of my nature that I didn't even know were there. But it's fun, it's good for the business, and I'm running with it, much to the chagrin of the NetBeans webmasters and our (much more qualified) web analysis team, who gets continually bombarded by my questions.

So what to expect? I'll be asking a lot of questions here, putting up the answers when I find them, doing a lot of complaining about "I can do this in this tool but can't in that tool" type of things. Hopefully you'll find it interesting. So stay tuned :-)




« July 2016