Use the Landing Pages Report to Isolate Self-Referrals

Landing Pages reportview at full size

With Full Referrer as a secondary dimension, the Landing Pages report can identify cross-domain issues.

In a recent post, we identified missing tracking code and incorrect cross-domain setup as the two primary causes of self-referrals manifesting in the Referrals report within Google Analytics.

We can take advantage of the Landing Pages report to isolate both of these issues. Specifically, if we apply Full Referral as a secondary dimension, and also apply the built-in Referral Traffic advanced segment just to temporarily hide other traffic mediums, you can see specifically where the breakdown is occurring.

• If the landing page and full referrer are on the same domain, the referring page is probably missing the Google Analytics tracking code. Easy fix: include the tracking code.

• If the landing page and full referrer are on different domains (main site and checkout site, as an example), you have probably not configured cross-domain tracking correctly. We’ll examine cross-domain setup in an upcoming post.

Please share this post:
Facebook Twitter Plusone Linkedin Email

Main Causes of Self-Referrals: Missing Tracking Code and Incorrect Cross-Domain Setup

20120714-self-referralsview at full size

The occurrence of your own website in the referrals report usually indicates missing tracking code or incorrect cross-domain setup.

It’s normal to see a few occurrences of your own website in your Referrals report. If a visitor waits more than 30 minutes between two page accesses on your site, the second pageview will count as new session, with medium as referral and source as your own domain.

If, however, the Referrals report shows a significant number of self-referrals, you’re probably dealing with either of the two issues below:

• One or more pages on your site are missing the Google Analyitcs tracking code.

• You are using the same tracking code on more than one domain and have not correctly cross-domain tracking (for a shopping cart that resides on a separate domain, as one example).

In upcoming posts, we’ll use the Landing Pages report to isolate both of these issues, and we’ll walk through correct setup for cross-domain tracking.

Please share this post:
Facebook Twitter Plusone Linkedin Email

Timeout Is 30 Minutes for Sessions But Only 5 for Real-Time Active User Count

Real-Time Active Usersview at full size

The Active Users count goes back five minutes.

A user drops out of the Active Users count in the Real-Time Overview report after five minutes of inactivity.

This calculation can be somewhat surprising, since a Google Analytics session terminates after 30 minutes of inactivity.

In either case, each hit that a visitor sends back to Google Analytics refreshes the visitor’s active status for another full five minutes and the session for another 30. Similarly, the expiration of the _ga cookie, which identifies a unique user, is refreshed for two years with each hit, and the campaign timeout is refreshed for six months. (You can override the default user and campaign timeouts in the property admin, but you’re probably fine to stick with the two-year and six-month defaults for most implementations.)

More about campaign timeout in an upcoming post.

Please share this post:
Facebook Twitter Plusone Linkedin Email

Video Play with Event Eliminates Bounce for the Session

A visitor begins a session on your home page, plays a video, and leaves. Does Google Analytics count that session as a bounce?

If you have used the default YouTube embed, the session will still count as a bounce. This is because Google Analytics records a visitor interaction only when it generates a hit.

If you have instead attached a Google Analytics event to a YouTube Player API embed, a video play will eliminate a bounce, since it will send a hit to the Google Analytics servers.

The same is true for a PDF download that you’re tracking as a virtual pageview, a Tweet link that you’re tracking as a social interaction, or a tracked Ecommerce transaction – all types of hits avoid a bounce for the session.

As a note, you can define an event as non-interaction. For instance, if a video begins automatically after 15 seconds on a page, you could still capture the video play as an event but opt to set the non-interaction parameter to true so that the video play in itself would not eliminate the bounce. For most events that you capture, however, a default interaction event is suitable.

In any case, it’s important to record all significant user interactions as some form of hit so your bounce rate and your overall Google Analytics data more accurately reflect user engagement.

Please share this post:
Facebook Twitter Plusone Linkedin Email

Abandonment Rate Is a Useful Metric, So Define Your Goals with Funnels

Abandonment Rateview at full size

Abandonment is calculated only for goals with funnels.

Everyone understands funnels, and everyone understands abandonment.

Since a significant part of our role as analytics is to communicate data – to others in our organization, to clients, and to ourselves – in a way that is clear and meaningful, we should define funnels on top of our Destination goals so Google Analytics can fully populate the user-friendly Funnel Visualization and Goal Flow reports and also calculate the emotionally resonant Abandonment Rate metric.

The more visual and emotional we can make our data, the more effectively we can demonstrate the need for, and the benefits of, optimization of key user interactions on our websites.

Please share this post:
Facebook Twitter Plusone Linkedin Email

Email Dashboards Monthly to Remind Colleagues, Managers, and Clients about Analytics

Email dashboardview at full size

Apart from the specific data points, emailed dashboards can be valuable as basic reminders.

Apart from the virtually endless combination of metrics and filters that you can configure into your dashboards, emailed dashboards can serve the very basic purpose of reminding executives, managers, colleagues, and clients that Google Analytics data is being collected and is available for further analysis at any time.

If your dashboard includes the most relevant metrics for your recipients, it is likely that they will periodically ask you to drill down into the data in ways that you, as the analyst, may not have thought of on your own. In short, they can help you ask the right questions for achieving actionable insights.

You can also email individual reports, but the inherent advantage of the dashboard is the variety of metrics it can present.

In any case, make sure to use the email feature in Google Analytics to keep your stakeholders engaged so they can help you focus your analysis on meaningful outcomes.

Please share this post:
Facebook Twitter Plusone Linkedin Email

Rather than a Single-Page Visit, Think of Bounce as a Single-Hit Session

Bounce Rateview at full size

Bounce is not defined by pageviews only.

While we may be tempted to define a bounce as a single-page visit, a much better definition is a single-hit session.

Let’s first examine hit. In Google Analytics, a hit corresponds to any data that is sent from your website (or app) to the Google Analytics servers, including:

• physical pageview (or screen view)

• virtual pageview

• event

• Ecommerce transaction

• tracked social interaction

If three users arrive on your home page and respectively play a video, download a PDF, and click through to your Twitter page, and you have those actions tracked as event, virtual pageview, and social, those sessions will not count as bounces, even if the users do not view any other physical pages.

As a note, hit in the context of Google Analytics doesn’t correspond to the broader meaning of hit in Web server parlance, which means any file request, such as HTML, image, JavaScript, or CSS. (As website optimizers, we want to remain aware of these types of hits as well – and minimize them to reduce download time for mobile sites – but this consideration is separate from Web analytics.)

Now, why session instead of visit?

If a user arrives on your site, views one page, speaks on the phone for 31 minutes, and then accesses another two pages, Google Analytics records two sessions for that same user, with the first session counting as a bounce. The visit consisted of three pageviews, but in the first session, there was only one pageview.

For these reasons, in defining bounce, hit is more accurate than page, and session is more accurate than visit.

It’s just one definition, but it encapsulates two important concepts in Google Analytics.

Please share this post:
Facebook Twitter Plusone Linkedin Email

View Traffic by Day of the Week

Day of the weekview at full size

You can display session by day of the week in a custom report.

Over the last six months, which day of the week saw the most visits to your site?

You won’t find an answer, at least not in aggregated numbers, within the standard Google Analytics reports, but we can easily display this data in a custom report.

First step is to create a custom report in Explorer format, with Day of Week Name defined as the primary dimension and Session (as well as other metrics, such as Bounce Rate, as needed) as the metric.

Once we display the custom report, we can apply Day of Week (as the numbers 0-6) as a secondary dimension so we can sort the report from first day of the week to last (Sunday-Friday). We can also apply alternate displays, such as Percentage or Performance, instead of the default Data display.

Don’t forget to take advantage of custom reports. They’re easy to use and available individually to each user, and they often allow you to display dimensions and metrics in ways that are not possible in the standard reports.

Please share this post:
Facebook Twitter Plusone Linkedin Email

Table Filters are Interpreted as Regular Expressions, So You Must “Escape” Question Marks

Table Filter Interpreted as Regular Expressionview at full size

Table filters are interpreted as regular expressions.

Though it’s not indicated anywhere in the interface, the table filter field in Google Analytics reports is interpreted as a regular expression (or “regex” for short).

In regex notation, there are two types of characters: literal characters and metacharacters. If you’re trying to filter your Pages report to display pages that contain article.aspx?id=, you must “escape” the question mark with a \ character so the ? acts as a literal character. Otherwise, the ? will be interpreted as metacharacter – specifically, a quantifier meaning zero or one of the previous character.

To further illustrate, if you enter article.aspx?id= into the filter field, you’ll match any pages that contain article.aspid= or article.aspxid=, but not article.aspx?id=, since the question mark is interpreted as a regular expression metacharacter and not as a literal. If you enter article.aspx\?id= into the filter field, you’ll match any page that contains article.aspx?id= since the question mark is now interpreted literally.

By escaping the question mark, we “escape” interpretation of the character following the \ as a regex metacharacter and allow it to act as a literal character.

Note as well that in proper regex notation, all literal . characters are also escaped so they’re not interpreted as the wildcard metacharacter that matches any single literal character. The filter article.aspx\?id= would also match pages that contained articlesaspx?id= and article-aspx?id= because the . is acting as a metacharacter. To restrict the match to article.aspx\?id=, you’d need to also escape the . as in article\.aspx\?id= and thereby force the . to be interpreted literally.

Don’t feel that you need to memorize regular expression notations. Know the principles, and download a cheat sheet at:

Please share this post:
Facebook Twitter Plusone Linkedin Email

Focus on Assists to Understand Where Your Conversions Are Really Coming From

Conversion Assistsview at full size

Assists can provide a much more complete understanding of conversion sources.

In a recent post about Multi-Channel Funnels, we examined the Top Conversion Paths report. Assisted Conversions, another report that Google Analytics provides under Multi-Channel Funnels, specifically indicates how often a channel is participating in Ecommerce or goal conversion assists vs. acting as the final channel for conversion.

In the Assisted / Last Click or Direct Conversions column, a value greater than 1 indicates that a channel is stronger with assists than with closes, and a value greater than one indicates a stronger closer.

Assists and closes are both good things, but if you were to review the Source/Medium list in the Goal Overview report, you would know only which channels were closing the deal.

As additional options within the Assisted Conversions report, you can:

• switch from Assisting Interactions Analysis to First Interaction Analysis

• change the lookback window from the 30-day default to as long as 90 days

• narrow your analysis to a single goal or Ecommerce transactions only

In any case, make sure that you’re referring to the Multi-Channel Funnel reports to understand where your most valuable traffic is really coming from.

Please share this post:
Facebook Twitter Plusone Linkedin Email