February 2010

Brief post today. First I wanted to tell everyone about Peashoot, a cool program that can follow link campaigns via Facebook or twitter. The best part is that, unlike HootSuite or Bit.ly which also can show stats from links on Twitter (or anywhere), Peashoot can connect into Google Analytics so that all information about how users interact with your web efforts can be collected together and shared. Here’s a more in depth review.

All of this can be better used to measure how the library is engaging patrons. Now engagement is kind of a trendy word these days. And, honestly, like most trendy phrases it is completely lacking in real substance. Well, analytics guru Avinash Kaushik briefly discusses how Google Analytics can be used to track engagement. What should we look for. What can be misleading.

The key insight in all of this, is that we have to define what engagement means for each campaign and each department. A teen library website will have far different criteria for engagement than a reading blog targeted towards adults. As usual, a successful program will have well defined goals and measures of success. Once those are defined, understanding the data is a snap.

So now that we’ve recognized the different user populations, we need to start following their unique traffic patterns. To do that we will use two tool: Custom Reports and Visitor Segmentation

Custom Reports

Since we already identified the key metrics for each visitor, we can make a custom report that will quickly give us the useful information in one form.

So we enter Google Analytics and select Custom Reporting

And we create a new report:

We’ll make a report for our local visitors, so we’ll drag and drop the following metrics: Entrances, Time on Page, Average Time on Page, Page Views

For the dimensions we’ll begin with City then we can ‘drill down’ further. This is where the power lies. We can look at people from Madison, WI (my town) and then click on that metric to see what the Landing Pages appear to be. This will only give us the top pages for people from Madison excluding users we don’t want. We can then drill down further to Exit Pages so for each Madison visitor, we can see the landing page and for each landing page we can see where the leave from (presumably because they found what they wanted or they were frustrated).

I’ll save this report as Local Searches and then when I run it and drill down two levels, I can see that the top landing page for users from Madison is the homepage (no surprise) and from that landing page I can see the top exit pages. Now I can begin to dig into what users are or are not finding.

Visitor Segmentation

We can generate a similar report using our key metrics for visitors. However, a more efficient way is to create a new visitor segmentation.

We’ll begin by selecting Advanced Segmentation

We’ll make a new segment and we’ll select City as the metric. This is a little loose and dirty and can be further enhanced with regular expressions or combination of metrics. However, we just want a quick division between those individuals that are local and those that are not.

After selecting City from Visitors we will segment out all visitors that are not from Madison. We’ll call this segment Not-UW (after University of Wisconsin).

After saving, it takes us back to our Advanced Segmentation window. Now if we click on our new segment we have full access to all reports that only involve individuals from outside of Madison.

And here they are. If we want to toggle back and forth we can select different segments from the Advanced Segments drop down at the top of the report.

In addition, we can have information from both segments (or many other default segments) combined into the same report .

When we apply, this will give us the report with different groups appearing in different colors:

Now that are different user groups are isolated, we can really begin to dig deep into what each user base wants and needs.

Not every visitor to your website is the same. They all have different goals, different expectations, different ranges of patience. Broadly, though, there are two important types of visitors that must be  considered separately. The first is a local visitor. This is someone who is directly connected to the system. Maybe they live in the service area, or are taking a class, or are somehow connected to the library system. A good rule is that these individuals will go to the website directly or through another local link (say a course page) because the have specific expectations of the site (that it has e-journals, or course reserves, or a online catalog).

The second visitor is a remote visitor. These individuals stumble upon the site through a web search or through links that are not directly related to them or their region. These visitors often are trying to find a specific item not a specific service.

Each of these broad visitor types will have very different behavior on the site. And each should be considered unique when developing a website.

Local Visitor

Expectations: They often have specific needs and expectations of services. Our job is to figure out what they want, where they expect to find it, and how we can best communicate and provide it.

Metrics: The metrics that can help us best learn this information are

1. Top Content: There are often many “home pages” and users may often go directly to the page that is most useful whether through personal bookmarks or links that they have received from someone else. In addition, this tells you where users spend their time. And you can find out how long (although this is deceptive. Did they spend time there because it was useful or confusing?).
2. Click Density: This is a nice guided tour to your site. This will show how users navigate, where they go and what they do. This is a great way to see what navigation paths work and are intuitive and which aren’t.
3. Time on Site/Page: This is similar to #1. Users that are local are using the site actively (hopefully). They are more likely to spend more time on different resources. Time on site/page helps determine which ones are useful (because they spend a lot of time using them) or confusing (because they spend a lot of time using them). The difference is hard to tease out and is best determined with surveys or other forms of qualitative study.

Remote Visitors

Expectations: A lot of remote visitors come to the site to find a single piece of information. They may want a electronic document, a phone number, information about a unique collection or other items. They do not necessarily expect to spend a lot of time using the site and working from it (although for rich digital collections, this may not be true).

Metrics The goal here is to find out a) what the user wanted b) if they could find it

1. Bounce Rate: This is a measure of those who went to the page and then left right away without viewing other pages. Remote visitors will push up the bounce rate very, very quickly. Why? Because they are likely browsing the web to find an answer. A quick glance at the page may tell them it is not what they want and off the go. However, if they go directly to a site that has the information they want and then leave immediately, it is still a bounce so you need to check bounce rate along side . .
2. Keywords: This is great because you can see exactly what the patron was expecting. Combine this with bounce rate and you have a good indication of whether or not they received what they wanted. By looking at the keywords, you can build an understanding of what users want from your site and how that can be incorporated.

Reports

This was a quick look at a few key metrics for different visitors. Next week, I’ll show how to make custom reports to help isolate the data about each of these visitors.

Photos: http://www.flickr.com/photos/88448902@N00/478136767/