Medium and Source

All GA sessions are recorded with at least two dimension values that describe acquisition: medium and source. As illustrated in Figure 7.1, Medium is the most general dimension value that GA uses to designate traffic acquisition, and Source is more specific. While source is sometimes used more generally in discussion to mean where your traffic is coming from, within the GA interface it refers specifically to the actual Source dimension.

By default—that is, without campaign tagging, which we discuss later in this chapter—GA records all traffic with one of three medium values:

• Referral: clickthroughs from any website that GA does not recognize as a search engine. The source dimension for referral traffic would be the website where the clickthrough originated, such as abc123.com or partnersite.pk.
• Organic: clickthroughs from one of the many websites that GA recognizes as a search engine and assigns organic as medium. The source dimension for organic traffic would be one of the many websites that GA recognizes as organic search engines, such as google, bing, or baidu.
• (None): direct entry of a URL into the browser (and, by default, many other types of traffic that you don’t want to appear as direct, as we discuss with campaign tagging later). All traffic recorded with (none)as the medium value is also recorded with (direct) as the source.

If you wanted to display the Source/Medium report aggregated by the more general medium values, you could change the primary dimension of the report to just Medium.

Referrals

As mentioned above, GA considers referrals to be clickthroughs from any website that it does not recognize as a search engine, and it assigns a medium value of referral to all such traffic (unless you have overwritten medium and source with campaign tags). Because the Referral Traffic report (labeled as Referrals in the left navigation) displays only sessions for which the medium dimension has been populated as referral, the report only lists Source as the primary dimension, as shown in Figure 7.2.

Full Referral Path

If you click into any of the sources listed in the Referrals report, you can see the specific page on which the link to your website was clicked, as shown in Figure 7.3.

Channels

The Acquisition Overview and Channels reports are organized by Default Channel Grouping. Designed as a more aggregated and user-friendly labeling than source/medium, the Default Channel Grouping contains nine channels, each of which is defined by medium, and in some by source as well.

As outlined in the “Default Channel Definitions” help article at https://support.google.com/analytics/answer/3297892?hl=en, GA offers the definition for the default channels:

• Direct
• Organic Search
• Referral
• Email
• Paid Search
• Other Advertising
• Social
• Display

Affiliates, an additional default channel that does not appear on the definition page, includes traffic for which medium value is affiliate.

It’s important to note that even though GA applies these default channel groupings to inbound traffic, in many cases it is up to you to provide the right medium and source campaign tags to inform GA how to group the traffic. To revisit a previous example, the Email channel is defined as Medium exactly matches email. This is quite straightforward, but if your inbound links from an email opened in Outlook aren’t campaign-tagged, the medium value will be recorded as (none), and the resulting sessions will be grouped as Direct.

Any sessions whose medium and source values don’t match any of the channel definitions appear in the Channels report as (Other).

Since we know where to look for the default channel definitions, there’s no guessing about which medium (and source) values to use as our campaign tags so traffic is grouped correctly in the Channels report.

As we discuss below, GA users with Edit access to a view also have complete flexibility to edit channel groupings as needed.

Treemaps Report

As a relatively new Acquisition report, the Treemaps report uses size and color to illustrate a simple yet ingenious comparison of two metrics. In Figure 7.4, the comparison of Sessions (size) to Ecommerce Conversion Rate (color) indicates underperformance of Display channel, which is especially problematic, since it’s a paid traffic channel.

Before you disable all your display campaigns, you can drill down to view a treemap that indicates Ecommerce Conversion Rate for specific display sources. You should also review the Multi-Channel Funnel reports as described in Chapter 9 to check if Display is providing Ecommerce conversion assists to other channels that are converting on last click.

In Acquisition ˃ AdWords, there is also a dedicated Treemap report for AdWords search campaigns.

Campaigns

In GA, campaigns consist of clickthroughs on links that you have tagged with special parameters—utm_medium, utm_source, and utm_campaign—that overwrite the default medium and source attribution and also populate the resulting sessions into the All Campaigns report. As a crucial task for effective GA, campaign tagging is examined in detail below.

Note that the following discussion relates to campaign tracking for traffic to your website as reported in a GA website property. Chapter 16, “Mobile App Measurement,” discusses campaign tracking for clickthroughs to Google Play or the iTunes Store for installation and also for clickthroughs directly to your app for reengagement.

Adding Campaign Parameters to Inbound Links

The process of campaign-tagging an inbound link is straightforward; you only need to add the following query parameters to the URL:

• utm_medium: the most general descriptor for the attribution; overwrites the default Medium dimension value
• utm_source: a more specific descriptor for the attribution; overwrites the default Source dimension value
• utm_campaign: the most specific descriptor for the attribution; this identifies the specific campaign and populates resulting clickthroughs into the All Campaigns report. As a convention, you can include the date at the beginning of the campaign name, especially for frequent, short-duration campaigns. (Technically, utm_campaign is not required, but you can consider it obligatory in practice.)

You can manually add these campaign parameters to your URL, but you’re instead advised to use a tool such as Google’s or E-Nor’s URL Builder (https://www.e-nor.com/tools/url-builder) to minimize errors.

Let’s use an email newsletter as an example for campaign tagging. The email that you send to your customer list each month contains news summaries and a link back to your main news page. Instead of linking with a nontagged URL, you’ll use the URL Builder for formatting a tagged link as shown in Figure 7.5.

By including the campaign parameters in the email link, we ensure that all resulting clickthroughs appear in Email as the Default Channel grouping, in the Source/Medium report as main-list/email, and in the campaigns report as 201605-newsletter.

Note also that you can use any of your pages—assuming that they all contain the GA tracking code—as your campaign URL. It is also routine to use the same landing page for multiple campaigns; the relationship over time is usually one-to-many.

As another example, let’s say that you’re tweeting about a product release. You could specify the following parameters:

• utm_medium: social
• utm_source: twitter.com (You could also use t.co, which is, strangely, the default source value for untagged clickthroughs from twitter.com.)
• utm_campaign: 20160601-product-release

Your formatted campaign URL would appear as:

   http://www.mysite.com/product-release?utm_source=twitter
   .com&utm_medium=social&utm_campaign=20160601-product-
   release

Table 7.1 lists many other types of traffic that require campaign parameters for accurate and specific attribution.

For the complete version of the campaign and channel matrix, visit www.e-nor.com/gabook.

Automated Pay-Per-Click Campaign Tracking

For your AdWords campaigns there is a much better option than manual campaign tracking. You should instead, in most instances, opt for AdWords Autotagging, which is easily configurable from GA as explained in Chapter 14, “Google Analytics Integrations — The Power of Together,” and reviewed below. Autotagging populates all five campaign parameters and a range of other useful dimensions listed in the “Benefits of Auto-Tagging ” GA help page. Note that the 25-character AdWords ad title is passed as utm_content. (To distinguish between two AdWords ads with the same title within GA, you’d need to apply AdWords Creative ID as a secondary dimension, or as a primary dimension in a custom report.)

Linking Google Analytics to AdWords to Import Cost Data and Enable Autotagging In the Property Admin, click Product Linking ˃ AdWords Linking to link GA to an AdWords account. This link will automatically import AdWords cost data into the GA views you select as part of the linking process, and will also, by default, enable Autotagging. Once you enable Autotagging, you should not add the manual GA parameters to the destination URLs of your AdWords ads. As a note, you can also import GA metrics into the AdWords interface, as discussed in Chapter 14.

Clicks Tab: AdWords Cost Data in Google Analytics If you have linked GA to AdWords as described above, the Clicks tab of the AdWords ˃ Accounts and AdWords ˃ Campaigns reports will display a range of cost-related metrics, including Cost, Cost per Click (CPC), Revenue per Click (RPC), and Return on Ad Spend (ROAS). RPC and ROAS are based on goal value and/or Ecommerce revenue. Keep in mind that ROAS does not take margin into account: if your CPC is 5USD and your margin per click is 4USD, you’re losing money, even though ROAS based on total revenue or goal value may appear wildly successful.

Linking Google Analytics to Multiple AdWords Accounts Note that if you link multiple AdWords accounts to the same GA views, each will be listed separately in the AdWords ˃ Accounts report. This may be useful if, for example, your organization is directly managing an AdWords account and you’re also working with an agency that’s driving traffic from a separate AdWords account.

Also, if you use an AdWords manager account (MCC), you can easily link all the accounts under MCC to Google Analytics to keep the AdWords accounts in sync in both systems. The MCC-level tracking is beneficial if you want to do cross-account conversion tracking and remarketing. For more details, see “Linking Multiple AdWords accounts to Google Analytics” in the Google AdWords help pages.

Auto-Tagging Bing Ads Bing Ads does not offer the same level of automated GA integration as AdWords in terms of additional dimensions and cost data, but you can enable Auto-tagging in the Bing Ads account settings to add utm_medium, utm_source, utm_campaign, and utm_term to campaign links (Figure 7.6).

Consistency Is Critical

Whether manual or automated, the mechanics of campaign tracking are quite straightforward. There is no technical difficulty in campaign tracking; the challenge lies rather in maintaining a consistent process and naming convention.

Naming Convention

There is no single campaign naming convention that we can point to as correct. What is most important is that you adopt a single naming convention and that you—and anyone else who is driving traffic and campaign-tagging inbound links—stick to it. Even if you remember to campaign tag for all scenarios outlined in Table 7.1, your analysis will be more difficult if you don’t maintain your naming convention, particularly for utm_medium and utm_source.

Figure 7.7 demonstrates unwanted fragmentation with Source selected as the primary dimension in the All Campaigns report. Rows 1, 2, and 5 indicate the same traffic source and should have appeared aggregated for more efficient reporting, as is also the case for rows 2 and 3. Inconsistent campaign tagging has made two traffic sources appear as five.

In the case of the Facebook campaigns, you may want to standardize on facebook.com as utm_source, since it is also the source value captured for nontagged traffic from Facebook. You can also control the case consistency by maintaining your naming convention, or you can apply a lowercase filter for each campaign dimension as described in Chapter 9.

Using the Same Campaign Name for Different Sources and Mediums

If you’re launching a campaign across different channels and websites, you have the option of using the same campaign name with different source and medium values, or you could choose to customize the campaign name per channel.

For example, if you’re announcing the appointment of your new COO on Twitter, on Facebook, and by email, you could opt to use 20160610-coo-tw, 20160610-coo-fb, and 20160610-coo-main-list respectively as the utm_campaign parameter, or you could just use 20160610-coo in all three instances.

In the first scenario, as shown in Figure 7.8, the All Campaigns report, with Campaign selected as the default primary dimension, would immediately distinguish between the different sources for that same campaign, but you’d lose the campaign-level aggregation benefit.

In the second scenario, as shown in Figure 7.9, the campaign value would be aggregated by default for all three sources, which means that you—and all others viewing the report—would need to add Source as a secondary dimension to view metrics for the campaign traffic from the different sources.

Either option has advantages and disadvantages, and both are perfectly valid. As stated earlier, there is not a single correct campaign-tag naming convention, but you should adopt your own convention and strive to maintain it with consistency.

Combining the Actual Referring Source with Campaign Parameters

As critical as campaign parameters are, their use does, in some instances, force a sacrifice: the loss of any actual referrer details, which are overwritten by the utm_source parameter. This issue might be especially troublesome in the case of press releases. If your press release is picked up by 50 online newspapers, you need to be able to aggregate all resulting traffic in GA, but you might also want to identify the publications individually.

With the help of a Custom JavaScript variable in Google Tag Manager, we can rewrite the page value passed to GA so that it combines the actual referring website with the campaign parameters when utm_medium=press-release (and otherwise leaves the URL unaltered in all other cases), so you achieve both the campaign aggregation and the referrer specificity that you need. You can then update your main GA pageview tracker as demonstrated in Figure 7.10 to set the campaignSource field to the value returned by the variable instead of the actual URL that appears in the browser.

You can download the JavaScript that you’ll need for the variable, and review the required tag and trigger modifications in detail at www.e-nor.com/gabook. An example resulting URL passed to GA appears below:

   http://www.mysite.com/updates.php?utm_source=pr-
   network-www.torontosun.com&utm_medium=press-
   release&utm_campaign=20160718-product-release.

Customize a Default Channel

In Figure 7.11, we saw that Feedburner clickthroughs were bucketed into the (Other) channel because they didn’t match any of the default channel definitions. If we want to include this traffic in Social, we can add an alternate match condition to the Social channel definition, as shown in Figure 7.12.

If you have Edit access to the view, you can modify the Default Channel Grouping by clicking Channel Grouping ˃ Channel Settings in the view admin. As with all changes that will affect the reporting for everyone accessing the view, try all modifications of the Default Channel Grouping in a test view before your working view(s).

Reorder a Channel

As discussed in Table 7.1, AdWords Autotagging adds cpc as the medium for Google Display Network (GDN) and remarketing traffic. Since Display appears lower than Paid Search in the Default Channel Grouping, GDN traffic from AdWords, by default, matches Paid Search first and the match with Display is short-circuited.

If instead you want your GDN traffic to count toward the Display channel, you need to drag Display ahead of Paid Search, as in Figure 7.12.

Define a New Channel

In addition to customizing an existing channel as outlined in the previous procedure, you can add a new channel to the Default Channel Grouping. At the top of Figure 7.12, we’re creating a channel to capture two different types of remarketing/retargeting traffic:

• Remarketing traffic from AdWords: Autotagging adds cpc as the medium to AdWords remarketing links, so resulting clickthroughs would, by default, match the Paid Search channel. We can instead define a new channel that would match all campaign names that contained remarketing. For this equation to work, we would certainly also need to include remarketing in the actual names of all remarketing campaigns administered through AdWords. (AdWords remarketing is discussed in Chapter 14.)
• Retargeting traffic from other networks: For other retargeting networks, such as AdRoll, we can manually add retargeting as utm_medium to the inbound links and include another match condition in our new channel definition.

Retargeting and remarketing mean the same thing; the latter is Google’s terminology. Match conditions for both appear in Figure 7.12.

Define a Custom Channel Grouping

In addition to creating or customizing individual channels, you can also create a custom Channel Grouping, that is, a customized set of channels. Custom Channel Groupings are probably not used as much as just custom channels within the Default Channel Grouping. Don’t feel compelled to create any New Channel Groupings, but be aware of the functionality if you need to take advantage.

Also, while changes to the Default Channel Grouping apply now-forward only, a new Channel Grouping applies retroactively. (The downside of the dynamic retroactivity is that new Channel Groupings are also subject to sampling, which we discuss in Chapter 10.) The new Channel Grouping does not replace the Default Channel Grouping; it serves as an alternative to the default, and you can readily switch between the two (or however many you create; the limit is 50).

As examples, you could define a new Channel Grouping with just two channels: Paid Traffic and Unpaid Traffic. Or you could define a Channel Grouping that breaks out specific referrers or social sources—again, configure the Channel Grouping however makes the most sense for a top-level traffic labeling scheme.

To define a custom Channel Grouping:

1. In the View Admin, click Channel Grouping and then + New Channel Grouping.
2. Name your channel grouping, for example, Paid and Unpaid Traffic.
3. Define your channels.

Note that you don’t need to redefine any of the predefined channels from scratch. To add Organic Search as a channel in your new Channel Grouping, set the rule to System Defined Channel – matches – Organic Search, for one example.

View-Level and Single-User Channel Groupings

In the procedure above, we began the process of creating a new Channel Grouping by clicking Channel Settings, shown in Figure 7.13, in the top section of the View Admin. To perform this action, you need to have Edit rights to the view, and, accordingly, this Channel Grouping will be accessible to everyone who accesses the view. The customizations to the Default Channel Grouping in Figure 7.12 also require Edit access, since they affect Channel reporting for everyone accessing the view. (We discuss access rights in Chapter 9.)

Each GA user with Read & Analyze access to the view also has the option of creating a new Custom Channel Grouping, as shown in Figure 7.14, that is accessible only to that user; it’s a private, user-level Channel Grouping, as indicated in Table 7.2. The GA user who created the Channel Grouping can share it with other GA users through Share Assets in the View Admin. Even more significantly, if the user who created the Channel Grouping has Edit rights, that user can “promote” the Channel Grouping to view level after first using it privately, thereby making it visible to all GA users who access the view.

	Default Channel Grouping	New Channel Grouping (View-Level)	New Channel Grouping (GA-User-Level)
Where do you configure?	Admin ˃ Channel Settings ˃ Channel Groupings	Admin ˃ Channel Settings ˃ Channel Groupings ˃ New Channel Grouping	Admin ˃ Custom Channel Groupings
Access rights required to configure	Edit	Edit	Read & Analyze
Who can see it?	All GA users who can access the view	All GA users who can access the view	Only the GA user who created the grouping
Does it apply retroactively?	No	Yes	Yes
Available in custom reports, in segments, as a secondary dimension, through the API, and in Analytics 360 custom tables (to avoid sampling)?	Yes	No	No

Channel Customizations in the Multi-Channel Funnel and Attribution Reports

The Channel Customizations described above apply not only to the Acquisition ˃ Overview and Acquisition ˃ All Traffic ˃ Channels report, but also to the Multi-Channel Funnel reports and the Attribution ˃ Model Comparison Tool, which we discuss in Chapter 9. For Analytics 360, customizations to the Default Channel Grouping apply to the Data-Driven Attribution modeling discussed in Chapter 18.

(not provided)

If you access the Campaigns ˃ Organic Keywords report, it’s highly likely that you’ll see (not provided) recorded as the keyword for the greatest number of organic sessions, as shown in Figure 7.15.

Branded versus Nonbranded Organic Traffic

The distinction between branded and nonbranded organic traffic is fundamental. Let’s say that your company is called Fallbreaker and that you manufacture and sell parachutes. If a searcher enters fallbreaker or fallbreaker.com into the search engine and then clicks through, you can consider the visit to be branded organic, also referred to as navigational organic. The searcher was already aware of your company, so you can celebrate your brand awareness, but you don’t want to congratulate your SEO team just yet.

If a searcher enters parachute into the search engine and fallbreaker.com ranks high enough to earn a clickthrough, you can consider this a true, nonbranded organic session and an SEO success.

Before October 2011, when the Google search engine began blocking keyword data from Web analytics, it was very easy to distinguish branded from nonbranded organic clickthrough performance in GA: each keyword appeared with performance metrics such as bounce and goal conversion rate, and you could quickly define a branded organic segment (including medium = organic, keyword containing F|fallbreaker) and a nonbranded organic segment (including medium = organic, excluding keyword containing F|fallbreaker) that you could then apply to any of the GA reports. (Segmentation is discussed in Chapter 10.)

These segments were particularly useful when applied to Goal and Ecommerce reports and typically revealed much higher conversion rates for branded versus nonbranded organic traffic. With nonbranded sessions isolated, you could measure and optimize true organic performance.

Google Image and Country-Specific Clickthroughs

By default, clickthroughs from Google image search results appear in GA with google.com/imgres as the source and referral as the medium. In Chapter 12, “Implementation Customizations,” we configure a view filter that rewrites these values as images.google.com and organic.

On a similar note, clickthroughs from country-specific versions of the Google search engine, such as google.co.jp or google.ng, appear in GA with google as the source. In Chapter 12, we configure our GA property to record specific source values for search engines with country-specific top-level domains.

Google Search Console

Google Search Console (formerly known as Google Webmaster Tools) can help us to address the previously discussed (not provided) issue, at least partially. As a tool separate from GA, Google Search Console provides reporting on visitor interactions with the Google search engine leading up to the clickthrough to your pages, and it also reports on Google search crawler (dubbed Googlebot) activity on your website.

Specifically, the Search Analytics report in Google Search Console will help you with (not provided) by displaying impressions, clickthroughs, clickthrough rate, and average positions for nonbranded and branded keywords searches in Google that resulted in organic clickthroughs to your website. (See Figure 7.16.)

Direct Traffic and Attribution Precedence

Let’s say that your company sells herbal supplements. On Tuesday, a Google search engine user searches for echinacea and clicks your AdWords ad (costing you 4USD, incidentally). The Source/Medium for the first sessions is google/cpc.

On Wednesday, remembering the 400mg echinacea supplements on your website, the same visitor enters your Web address directly into the browser (and purchases the echinacea this time). So the second-session Source/Medium would be (direct)/(none)—correct?

Actually, in most GA reports, the second session will also be attributed to AdWords and will appear with same Source/Medium values and channel grouping as the first session.

This makes sense. The second visit, though technically direct, is clearly attributable to the first visit, which was driven by AdWords. As long as the second session takes place on the same device and in the same browser, the user has not deleted cookies, and the browser in the first session was not set to private/incognito mode—private/incognito browsing automatically deletes cookies after a session—the second session will still count as google/cpc.

Other Source/Medium values have equal precedence and therefore overwrite each other, as demonstrated in Figure 7.22. For instance, if the second session above had been organic, a referral, or a campaign clickthrough, any of these sources would have appeared in GA for the second session and not google/cpc, the idea being that the nondirect returning source deserves credit—at least to some degree—for the returning session.

In Traffic Info ˃ Session Settings within the Property admin, you can see that Campaign Timeout defaults to six months. This determines how far back the GA reports will look back for a direct visit to pull in a more specific Source/Medium from a previous visit. You can easily change the Campaign Timeout if you choose, but it’s generally recommended to keep the setting at the six-month default. (See Figure 7.23.)

Direct Sessions in the Multi-Channel Funnel Reports

As mentioned earlier in the chapter, the Multi-Channel Funnel reports display Source/Medium details for all sessions during which a conversion occurred or that were followed by a returning session in which a conversion occurred. Unlike the Acquisition reports, the Multi-Channel Funnel reports assign equal precedence to direct traffic relative to other sources and do not therefore repeat Source/Medium from previous sessions for returning direct sessions. The Multi-Channel Funnel reports are discussed in greater detail in Chapter 8.