Measuring User Engagement

Have you ever been in a marketing meeting where someone stood up and said, “Hey, we need to be advertising on ABC.com because it’s all the rave!” While, the awesomeness of the site might be a justifiable reason for some of us to spend $10,000/mo on a campaign on ABC.com, the rest of us what to know how that site will perform.

Or on the flip side, we just completed a campaign where we advertised on two different sites and it’s time to decided if we will continue advertising on the site or change tactics.

For either reach we can provide statics to gauge the level of visitor engagement from said sites. The other day I was reading the Google Analytics blog and came across a posting about an API update. http://analytics.blogspot.com/2009/12/api-python-client-library-updated.html

The post describes the use of the Analytics API Python Client library and a sample application they’ve created to test the library. The application creates segments to pull specific referrals from select referring domains. Then, it takes five calculations and corresponds metrics to compare the performance of those segments.

I decided it would be a great experiment to test out the application as well. With the help of a programmer, we were able to create the pretty chart as displayed on the Google Analytics Blog. Exciting.

Since I’m not a programmer and had some time to spare I decided to create the same chart using Google Analytics and Excel. I took the same steps, Segmented my referring audience. In this example I created two Advanced Segments to capture all referrals from two separate campaigns. Then, from the Google Analytics Dashboard, I copied the following five metrics into excel for each segment. Created a few charts, formatted a bit and BAM. A chart that compares visitor engagement between two referrers.

My chart might not be as pretty nor as quick to create from the API application, but it does the trick for now. I’ve picked out a few Java API applications I want to play with next when I have a few minutes.

Ok, so now we have this fancy chart and I’ve explained how to create it without having to be a programmer who understand Python.

What now, what does it all mean? Well… For starters, we can see that traffic from Campaign #2 is more engaged compared to traffic from Campaign #1. Visitors from Campaign #2 view more pages per visit, on average they spend more time on the site and they return more frequently. The Bounce Rate is higher on Campaign #2 compared to Campaign #1 but compared to the site average, Campaign #2 is lower. If I had $10,000 to spend, where would I spend it? I would split it between both campaigns. Maybe a 30/70 split, depending on the site’s objectives. Campaign #1 drives and retains a nice chunk of traffic to the site.

This fun new chart provided us with a great deal of insight, not only from an evaluation standpoint but also to look at potential marketing opportunities.

Using data to make marketing decisions. Interesting concept.