Today Google has introduced a new feature which should be available to every Analytics user - well it is available here in Australia anyway, which made me think it should be everywhere!

The new feature is called Weighted Sort, and comes into play when you click on any metric in a Google Analytics table, with the intention to sort ascending or descending. Without clicking to sort, you won't see it.

The purpose of it is to help you sort out those metrics which are worth looking at, and those which aren't.

For the blog below, I want to look at Bounce Rate. When I click on Bounce Rate to see the worst offending pages, I get all these pages, which have had not many views. What would be more useful, is to see pages which had a more significant number of views, and which also had bad bounce rates. This is where you use Weighted Sort - as shown below;



See, the page ranked at the top of this table doesn't necessarily have the worst bounce rate, but it has a whole lot of visits, so it is pretty important.

You can only use this new functionality with some of the calculated metrics - e.g. bounce rate and % new visits.

I am writing this piece just because the instructions provided by Google are hidden away among the instructions for various other things, and also because it stops (in my opinion) just short of what I wanted them to explain.

So, I hope this is a helpful and comprehensive explanation on how to track subdomain traffic when using Google Analytics.

In my example, Suzy's shoe shop www.suzyshoes.com, has some subdomains - www.stiletto.suzyshoes.com and www.slipper.suzyshoes.com.

Each of these subdomains has a contact page, as does the main domain;

www.suzyshoes.com/contact
www.stiletto.suzyshoes.com/contact
www.slipper.suzyshoes.com/contact

The problem with this is that in Google analytics, reports show only the trailing slash page. E.g. if each of these pages got a visit, then in Google Analytics, Suzy will see 3 visits to the page "/contact", but this is in fact 3 different pages

To improve on this reporting, Suzy needs to make changes both to her Google analytics code, and her Google analytics settings.

Changing Your Google Analytics Code
First step, the code changes. You need to add a snippet of code into the existing code. Note here I am using the old code, not the new asynchronous one. If you are using the new asynchronous code, check out the instructions here. So, below is a bit of the code which is on Suzy's site already, and the red line is the new bit she needed to insert to track the subdomains separately.

<script type="text/javascript">

try {
	var pageTracker = _gat._getTracker("UA-XXXXXXX-X");
 pageTracker._setDomainName(".suzyshoes.com");
	pageTracker._trackPageview();
}catch(err){}
</script>

Changing Your Google Analytics Setup
Now that this tracking is in place, you need to make some changes in your GA settings.

First, make duplicate profiles of the one you are working on. So, Suzy will leave her existing one as is, and also make three more profiles;
1. Excluding subdomains
2. Stiletto Subdomain
3. Slipper Subdomain.

Then, in each of these three new profiles, she will implement the following filter;
Filter Type: Custom filter -> Advanced
Field A -> Extract A: Hostname -> (.*)
Field B -> Extract B: Request URI -> (.*)
Output To -> Constructor: Request URI    /$A1$B1
Field A Required: Yes
Field B Required: No
Override Output Field: Yes
Case Sensitive: No

The result of this is that in each of these profiles, her page reports will now show the visits to her contact page like this;

.suzyshoes.com/contact
.stiletto.suzyshoes.com/contact
.slipper.suzyshoes.com/contact

So she can see exactly what contact pages got the visits.

Now, the last step is to include/exclude the traffic in each of the different profiles, so as to make them just for one particular subdirectory.

You need to install a filter on each of the profiles, which looks something like this;

Filter Type: Custom filter -> Include
FIlter Field: Hostname
Filter  Pattern:  ^stiletto\.suzyshoes\.com$
Case Sensitive: No

Suzy would then repeat this for the slipper profile, but for the other profile (which is excluding the two subdomains), she would implement two filters, making each of them 'exclude' instead of 'include'.

For more information or specific help, call us about our web analytics consulting.

Since the introduction earlier this year of Advanced Segments in Google analytics, I have found a lot of great uses for them.

Some examples include:

 

These segments can be applied to nearly any part of the Google analytics reporting suite, and can even be shown graphically on the main graph.

Since the introduction of Advanced Segments, there has been some questioning of whether this would mean you don't need profiles and filters any more. These were the original ways to do 'advanced segmenting'. You would make a duplicate profile of your account for every different 'segment' you wanted to follow, and Google allowed you up to 50 profiles.

Google Profiles are the second rung under your 'accounts' in your Google Analytics account. A profile can be a new web address, OR it can be the same web address as one you are already tracking but with filters in place to manipulate the data. For example:

Now that there is advanced segmenting within the reporting interface, do you need to do extra profiles and filters? Here is a quick list of differences between them:

Profiles with Filters

• Cannot look at data retrospectively
• Can easily be shared with any user, you just grant them access
• Slightly more complicated to establish
• They may be more accurate. Filters are based on page views, while advanced segments are based on visitors.  So, for example, Advanced segments will ignore all page views connected to a user who has seen restricted pages in their visit, whereas Profiles with filters will acknowledge that visitors behaviour except for the visits to the restricted pages.

Advanced Segments

• Allow you to look at data retrospectively
• Are much simpler to implement, with a better user interface
• Allow more flexibility, allowing more restrictions/requirements per segment.
• While you can't easily allow access to another user, they do have a 'share' URL which you can send to anyone, which will allow them to implement the exact same segment with a click of a button.

I continue to use Profiles and filters for big segments of site traffic I want to follow, things that are important and will continue to be important to the business.

If I want to do retrospective segmenting though, I can use the Advanced Segments function.

A few things to remember when creating profiles with filters

1. Keep at least one version of the original
2. When comparing profile traffic, make sure they are comparing equally, e.g. if you exclude traffic from your business's IP Address from one, then you need to do that to any you are comparing it to.
3. Profiles will only track traffic from the moment they are created, so try and think of ones you want to create as early as possible.
4. If you need to compare certain performance of different profiles, ensure they have the same goals set up.

When I opened up my Google Analytics today I noticed that I have been given access to the Beta for the new Google Analytics alerts - shown above in the LHS menu (under "Intelligence") and below the traffic graph.

How it works is, if Google analytics detects a change in the pattern of your site traffic, or other metrics, it can send you an email to let you know.  This way, you don't have to log in to keep checking, you can rest assured that if something important happens, you will know.

Below is a screen shot of how you create the alerts - first you choose the thing you want to monitor (in this case a keyword), then you choose the metric (in this case I chose visits), then you choose a condition - you might want to make it over a certain amount (your average, for example), or a % increase or decrease could trigger the alert.



This way, for example, I could get an email alert sent to me whenever the number of visits for the term "digital media" exceeds 200 or falls below 50.

For your business, maybe you want to know when your 'Melbourne' campaign has started working, so you set up a Google analytics alert to trigger visits from Melbourne increasing above average. 

Maybe your site relies heavily on traffic from a certain referrer or source - you can make an alert for when traffic from that important source dips below a certain level.

You can monitor almost any Google metric, and I think nearly every business could find a use for them.


.

Recently I have been harping on about the ups and downs I have experienced in my blogging due to reactionary posts I have written r.e. news articles.  I feel particularly ashamed of this behaviour because of the general fury I get, both at previous jobs when managers took my analysis and ran away with it creating a completely differend context, and when I discover the media is trying to swizzle me with their beat ups.

People try to do this to news/data for their own reasons, be it to make themselves look good, scare people into a certain action or to just create some buzz. But having had the title ‘analyst’ in nearly every job I have had since leaving university, it annoys me greatly, as it negates a lot of the work we analysts do.

To stop people doing this to YOUR data in the future, I have put together my little list of Data Contexts – how to try and ensure people don’t misinterpret your data.

If you provide this information alongside your data, you can try and minimise the ways people can misinterpret it. Just remember, if you put too many words/graphs/tables around your base results, you run the risk of people not reading, which encourages misinterpretation. So keep it minimal, but try to get the most important bits across to your audience (where possible):

1. Compare to the previous month/week/quarter/year. Put it in temporal context – is it normal or abnormal?
2. Seasonality effects – if it is abnormal, is it due to changing seasons? Holidays?
3. Compare to the industry/sector, town/country market – is it abnormal now? Or are you following the rest of the market? Compare using competitive intelligence tools like Hitwise or Google Trends
4. Segment your data – can you isolate where the changes are? Are your increases in sales due solely to one product? One segment of customers? One area? Do the sales translate to profits? Drill down into your data.
5. Compare with offline data. E.g your offline marketing campaigns; your server functionality;  have changes to your website resulted in code being dropped?;has someone just nay sayed your brand online?;has your competitor just gone bust?; has the Government just sanctioned/outlawed your product? Try and think of possible reasons, but remember to temper these with common sense, as they are pure headline fodder.

Ok, maybe that is enough for now- like I said, too much context and people lose interest. But seriously, don’t just be handing over a number in a “What is the meaning of life?” kind of exchange. You know a lot more about that number you are handing over, so let people know about it. Don’t give people an easy headline, it will only bring you headaches later when they then ask you to support their crazy assumptions. (”Ok Tracy, I just need the data to back up what I told them about our clients living longer than competitors”).