Making Sense of the Data, Part 2

This post is Part Two (see part 1 here) of a series adapted from my presentation at the 2011 HOW Interactive Design Conference in San Francisco. I included a variety of slide transitions in my talk, so in some cases below, I’ve consolidated slide groups into animated GIFs. Keep your eye out for those so you don’t miss any details.

Last time, I debunked our typical approach to web measurement and urged for a more holistic, functional one. I ended by mentioning what I think the two most important things about measurement are: (1) There are no independently meaningful metrics, and (2) Anything can be a source of data. So I’d like to pick up there, with Thing #2, by digging in to one particular source of web data—Google Analytics. We’ll see how Thing #1—that there are no independently meaningful metrics—really applies here.

Gathering Data from Google Analytics

Google Analytics is a big world. I just want to scratch the surface with you here and provide a basic approach that you can use on a regular basis. Once you’ve been using it for a while, you can probably stick to a weekly to biweekly schedule.

There are four metrics worth looking at regularly. They are:

1. Traffic Sources
2. Top Content
3. Bounce Rate
4. Goals (conversions)

But remember, it’s important that we don’t look at these individual metrics independently. They are meaningfully interconnected!

If you’ve never used Google Analytics, here’s the first thing you’ll see once you’ve logged in:

This particular view (above) of the Google Analytics dashboard comes from a client’s accounts—they offer used car loans and see a large volume of traffic to their site. The Google Analytics dashboard will give you an overview of a website’s traffic over the last month by default (you can configure it for any date range). Lots to see here, of course, but once I’ve glanced at it, assuming there are no major dips in traffic, I usually head right over to check in on Traffic Sources (below).

Here, I’m really just interested in getting a sense for where people are coming from. For this particular website, I can see that most people are coming in directly, others are coming in through search engines, and some from other places like Craigslist. But now that I’m digging in, I’m curious: Which sources are bringing in users that convert in some way?

I can figure that out by applying an advanced segment to this report.

I’m going to walk you through the flow of applying this advanced segment by using the animated GIF above.

An advanced segment is a way of filtering any report on the basis of some other metric. Google Analytics lets you create as many as you want, and configure them in just about any way you like. In this case, I want to put my Traffic Sources report through a simple filter that just identifies goal completions (Step #2 above).

Once I’ve applied that segment, I can now see some new information in my Traffic Sources report (Step #3). First, I can see a new orange line along with the traffic graph at the top. This is telling me how many goals have been hit from day to day—obviously, less than the overall traffic. Below that (Step #4), I can also see each source with it’s own goal conversion rate in green. This helps me value traffic sources. Traffic is exciting, but it’s only meaningful once it’s valued by some other metric. Conversions are probably the best way to value traffic.

The second report I routinely look at is the Top Content (above). This shows me which pages on a website are getting the most views, how long users are spending on them, and which pages have the highest bounce rate.

Bounce rate, by the way, is one of those metrics that people get hung up on all the time. Here’s a simple definition for what bounce rate is: The bounce rate is the percentage of traffic that landed on a page that then left the website without looking at anything else. You can average that for all the pages of a website and come up with a site-wide percentage, which is often what most people are focused on, but I think it makes much more sense to look at bounce rate for individual page performance instead. Let’s explore bounce rate on one particular entry listed in the Top Content report:

I’ve isolated one page from the list of top pages (above). The Top Content report tells me that this page has received over 15,000 unique views. Just over to the right a bit, it also tells me that the bounce rate for this page is 6.49%—amazingly low, by the way. So the question I’m left with is, how many users bounced? What most people would do here is apply the bounce rate to the number of unique views. It’s a simple math problem: 15,615 x .0649, which produces 1,013. But is that the right answer? No. The correct answer is 274.

Wait, what?

Remember, the bounce rate only applies to traffic that landed on a page. All the other unique views a page gets are from users that have landed on other pages and eventually made their way to it. So, to figure out how many people actually bounced from a page, we need to look at the Top Landing Pages report instead (below).

Instead of telling me about all the views a page receives, the Top Landing Pages report (above) only shows me how many views were first views—the first page a visitor saw on a website. If I look at the same page from the other report, I can finally get to the bottom of this bounce rate thing.

Since I’ve isolated that page (above), I can see that 4,225 people landed on this page (less than a third of its overall views), and only 274 of those people bounced. 274 is 6.49% of 4,225. The Top Content report didn’t make that very clear, did it?

Digging in to bounce rate for a particular page reveals just one example of how there are no independently meaningful metrics, and how you have to keep your big-picture questions in mind when navigating analytics reports. Otherwise, you’re vulnerable to what I call “analytics myopia,” which is what gives data the power to mislead rather than empower. So, always evaluate your Traffic Sources and Top Content in light of Bounce Rate and Conversions.

Have we had enough numbery stuff for one day? Does your brain feel a little bit like a jumble of tinker toys? Yeah, mine too.

So, we’ll pick up next time with the final entry on making sense of the data, in which I’ll show you how to gather data from your most important source: Real, live users.

 


Resources Recommended by Imprint

Related Articles:

ADD A COMMENT

2 COMMENTS

  1. Pingback: Tweet-Parade (no.4 Jan 2012) | gonzoblog.nl