2016-09-28

Meeting Information

Date: 9/28/2016
Time: 2pm
Location: JC311D
Streaming Link: https://bluejeans.com/834348332

Agenda

Mason Analytics Update

New Sites in Roll-Up
Hostname Business Unit
124.137.58.12
www.cssr.gmu.edu CHSS
ws.sharethis.com
volgenau.gmu.edu.googleweblight.com VSE
spmoauth.gmu.edu
m.baidu.com
gmulaw.org LAW
fanyi.myyoudao.com
drupalvm.dev
www.dvydcswnmmrs.fr
business.gmu.edu.googleweblight.com Business
housing.preprod.gmu.edu
seor.gmu.edu VSE
cvpa-dev.preprod.gmu.edu
Configuration Changes

Referral Exclusion List

2016-09-21: Added “gmu.edu” to referral exclusion list to prevent new sessions from being created when moving between subdomains.

Without this entry in the referral exclusion list, GA was creating a referral when moving between websites and therefore a new session.

Adding “gmu.edu” takes care of all subdomains under “gmu.edu”. Traffic moving from one gmu.edu subdomain to another will no longer count as a referral.

Annotations noting this change have been added to roll-up views.

Questions for Follow-Up

What are people clicking on when they land on a program page (from Mason referral for example)?

Test Case: What do people click on from the Accounting, BS program page (depending on how they got there)?

Account: School of Business
Property: business,gmu.edu
View: business.gmu.edu

Option 1: Behavior Flow

Behavior -> Behavior Flow -> Set specific landing page filter -> Create segment for referrals from Mason Core (Referral Traffic from Mason Core)

Note: once you add a segment to the behavior flow report, you can switch between viewing different segments using the buttons at the top. (Unlike other reports, multiple segments cannot be viewed simultaneously on this report.)

For comparison, add the Direct Traffic segment.

Note the differences in 1st interaction page between segments.

Option 2: In-Page Analytics

Behavior -> In-Page Analytics

I recommend using the Page Analytics Chrome Extension.

While this is very interesting in theory, I have often had issues getting reliable data.

Note that multiple links to the same page can not (usually) be differentiated.

Despite these issues, the summary info at the top is useful

Where is traffic coming from (for program pages)?

Account: School of Business
Property: business,gmu.edu
View: business.gmu.edu

Option 2: Custom Report -> “Business: UG Program Page > Source/Medium” -> Switch to pie chart view -> Drill-down on landing page for program

Review custom report configuration.

Compare referring source and medium for selected pages.

Data Sampling

Data sampling can affect your reports!

Example 1:

View: gmu: roll-up 2.0 – all traffic (2015-12-08)
Custom report: Dimension: Hostname > Page
Date Range: Jan 1, 2016 – Sept 21, 2016

See sampling notice, number of hostnames.

Change sampling control slider to highest precision.

See sampling notice, number of hostnames.

Change sampling control slider back to middle.

Example 2:

View: gmu: roll-up 2.0 – all traffic (2015-12-08)
Custom report: Campaign: Admissions UG > Campaign > Landing Page
Date Range: Jul 1, 2016 – Sept 21, 2016

Drill-down to Source / Medium -> “honors” campaign.

See sampling notice, number of landing pages.

Change sampling control slider to highest precision.

See sampling notice, number of landing pages.

Facebook Clicks Versus Sessions

A couple users have noticed discrepancies between the number of clicks that Facebook reports on ads versus the number of sessions that Google Analytics reports form the same ads.

A few contributing factors:

  • Clicks and Sessions are different metrics – Facebook counts clicks, Google counts sessions. The same user could click the ad multiple times in the same session.
  • Browser configuration issues – For Google Analytics to capture data, the user’s browser/device and security settings must support cookies, JavaScript, and images. If any of these are disabled, Analytics may not be able to record a session, while Facebook would still be able to count clicks.
  • User Behavior – Since GA depends on Javascript, it is possible that users could click away from a page before the user’s browser has a chance to send the analytics data, especially for sites that take longer to load.
  • Data Sampling – GA may be sampling your data in reports which could result in a lower-than-actual number of sessions.
  • Bot and spider filtering. – Depending on your GA implementation, GA may be filtering-out ‘questionable’ traffic from your data. Does Facebook filter out ‘questionable’ traffic?

Nicole: “Basically, Facebook says don’t worry about it: our numbers are correct and we wouldn’t lie. And a big chunk of their advertisers are saying, the discrepancies in reporting are too big to be explained by clicks vs sessions.”

Facebook has a financial incentive to count every click it can, while GA has an incentive to provide as reliable data as possible.

More information:
Wall Street Journal: Facebook Overestimated Key Video Metric for Two Years
Why AdWords Clicks and Analytics Sessions don’t match in your reports
Data discrepancies between AdWords and Analytics

Users, Sessions, and Hits

What is a ‘user’?

A user is a unique visitor to your website.

What is a ‘session’?

“A session is a group of interactions that take place on your website within a given time frame … a single session can contain multiple screen or page views, events, social interactions, and ecommerce transactions.”

How do sessions end?

  • Time – A session will end after 30 minutes of inactivity (but this timeframe can be changed). Inactivity means no GA hits of any kind within 30 minutes.
  • Date – A session will end at midnight, regardless of user activity. (Note: ‘midnight’ is determined by your view timezone settings.)
  • Campaign – Every time a user’s campaign source changes, Google Analytics opens a new session (regardless of user activity). (Note: a source of “(direct)” will not create a new session. – “Direct as a campaign source never overrides an existing known campaign source like a search engine, referral, or campaign source would.”)

More info: How a session is defined in Analytics

What is a ‘hit’?

A hit is an interaction with your website that results in data being sent to GA.

Common hit types include:

  • Page – pageviews
  • Event – actions a user takes on your site, other than viewing pages (examples: clicking on a tab on the page, scrolling, opening expanding sections, clikcing on videos, etc.)
    Hits can be categorized as ‘interaction’ or ‘non-interaction’ hits. Interaction hits will factor into your bounce rate, non-interaction hits will not.
  • E-Commerce – online transactions
  • Social Interaction – social media activity

Campaign Tagging Coordination

You can use campaign parameters to identify how traffic reaches your website. Campaign information is appended to your HTML links to provide additional, customized data.

When someone follows a link to your website, GA records the hits and populates the standard GA dimensions/metrics for the hit in its database.

Campaign tagging basically give you an easy, self-service way to customize the data loading into in a few key dimensions which are used in your GA reports.

Campaign information is completely customizable (which is both good and bad).

Campaign tags include:

  • Medium – (Required) What type of traffic is it? (examples: email, social, display, print, etc.)
  • Source – (Required) Where did this traffic come from? (examples: facebook, radius, newsletter, etc.)
  • Campaign – (Required) The name of your marketing campaign, or other grouping of communications
  • Content – Typically used to differentiate ad content, but can be customized. (Note: this parameter is listed as “Ad Content” in the Google Analytics user interface.)
  • Term – Typically reserved for paid advertising keywords, but can be customized

Key Points

  • Google analytics is case-sensitive. Therefore, for consistency, it is recommended that your campaign parameter values should always be lowercase.
  • All links used that you intend to track should include campaign parameters. Links not using these parameters will be indistinguishable from normal web traffic (direct).
  • Your campaign parameters should be decided on thoughtfully and used consistently. If they are used with different capitalization, spacing, spelling, etc. they will appear as different campaigns in GA.
  • I would recommend putting together an excel spreadsheet of all of your campaign-tagged links.

It is very important that you define what campaign tag values you will be using, and stick with them. Forgetting to use the link parameters, using them inconsistently, or mistyping them will affect your data and your ability to track your campaigns. For this reason, I would suggest adding as a standard procedure that every communication and all links within it are reviewed before the communication is sent out to make sure that all links are tagged appropriately.

Note: all campaigns are feeding to the roll-up, so it would be a good idea to coordinate. Although you can filter the campaign reports for your specific parameters, the default reports don’t do this, so identically-named campaigns will be aggregated (example: F16, Spring2017, etc.).

Examples

Enrollment Management – Emails

Medium – “email”
Source – “admissions connect” (for UG) or “admissions radius” (for GR)
Campaign – US regions (and “honors”) – we are dividing our communication plans by region (and “honors”), so our campaign names reflect that
Content – used to indicate the phase in the admissions process (“inquiry”,”applied”,”admitted”,etc.)
Term – not used

This schema enables us to slice out campaign data by region or admissions-process phase.

CHSS – Emails

Medium – “Email”
Source – “CHSS Radius”
Campaign – used to indicate the phase in the admissions process (“CHSS Inquiry”,”CHSS App In Progress”,”CHSS Admitted”, “CHSS Intent to Enroll”)
Content – reflects the specific email in the campaign (examples: “ANTHRO Day 15”, “ART HIST Day 30”, etc.)
Term – not used

As they are not segmenting comunications geographically, they are able to use the admissions-process phase as the campaign, which frees-up the content field to note the specific email.

I would recommend lowercasing everything.

OCM

Medium – “Post”, “tweet”, “video”, “news article”, “graphic”, “general-info”
Source – “Facebook”, “Twitter”, “Instagram”, “LinkedIn”, “YouTube”, “Google+”, “Pinterest”, “Reddit”, “Wikipedia”, “Tumblr”, “Snapchat”
Campaign – FY_Source_Post-Date (Example: FY17_facebook_071916)
Content – Descriptor of item (e.g. author’s name, details of graphic/image, name of video, 3-4 words max)
Term – not used

I would recommend lowercasing everything and changing the date format in the campaign to YYYYMMDD, to allow filtering by specific years/months (i.e. campaign begins with “201607” would catch all communications in July 2016).

Links/Resources

Google Analytics Demo Account

https://support.google.com/analytics/answer/6367342