You’ve used Google Analytics for years, but how deep have you searched into its hidden, labyrinthian corners? In this session, you’ll get introduced to the reports, dimensions, and other features you never knew would be so valuable. From using service provider names to identify your audience, to site search terms to inform your content strategy; how to share dashboards and automate your emailed reports; and best practices for filtering, campaigns, and event tracking. We’ll end with a demo of using the Google Analytics API with Google Spreadsheets and embedded charts to build your own custom reports.
We’ll be using case studies of Google Analytics implementations from real nonprofit and government websites, campaigns, and mobile apps to illustrate all these best practices. You will walk out of this session with a list of immediate next steps for enhancing your Google Analytics implementation, a wealth of knowledge about how you can push this tool to help you better measure your organization’s digital goals, and new techniques for reporting website data to your stakeholders.
2. google
analytics
@devonvsmith
I’m Devon Smith
Founded the analytics practice at Threespot
Digital agency in DC, exclusively serving nonprofits, foundations, & government agencies
I’m using their data, but hopefully anonymized
Focus on *communications* data in large part
3. slideshare.net/devonvsmith
05: secrets
40: Reports
20: data
20: process
05: beyond GA
@devonvsmith
90 minutes together to look at:
5 secrets, 7 reports, 5 front-end code implementations, 6 admin customization, and 1 script
Assume you have some knowledge about GA
Happy to stay after to discuss Q&A, or reach out to me on Twitter, also posted on Slideshare
4. step 1
ask questions &
test hypotheses
Don’t just open GA and click around to see what you can find
Develop a list of questions you want to get answered,
or a list of hypotheses you want to prove/disprove
Look for the data to support those, and lead to new questions
5. step 2
Use less data
Sounds counter-intuitive, but look at the leaf, not the forest
Ignore everything about your site except for one data point
Advanced segments, advanced searches, secondary dimensions
6. step 3
add better data
Don’t rely only on the data that GA gives you by default
Front-end code for event tracking, social actions, linking accounts, customize admin section
7. step 4
extract data
Process: use the tools built into GA to export data in various forms,
so that you can manipulate it in excel/google spreadsheets
8. step 5
add insights
Google Analytics only tells you what is happening,
not why, or how you should change your approach.
Every report you deliver to stakeholders using GA data should include a
“Here’s what we’re going to do differently next time around” section.
9. Better
Reports
@devonvsmith
GA’s key reports are: Who —> Audience; How —> Acquisition; What —> Behavior
Assumes no code knowledge, just manipulating reports
7 reports I use most often
10. @devonvsmith
who (actually)
visits our site?
The very first thing I look at for any new project
Often when I ask, who does visit, or who do you want to visit, orgs have fairly generic answers:
Funders, Government, Media, Peers
But there are huge variabilities within any of those segments
So let’s dive deeper…within the Government, do you care about local, state, federal? In certain areas? Agencies?
Looking at what type of content on your site? Are they finding it?
Network report answers these questions
And is probably most under-used report in GA
11. List of service providers - the name of the internet service provider of the person browsing your site
Top of the list will always be mobile phone carriers, generic home internet providers
But you’re looking for the visitors browsing at work
Use an advanced segment to narrow down this list
12. I’ll show you how to create a segment later in the workshop,
for now, these are segments I’ve created based on some current clients
Choose: Federal Government
13. [share link to segment]
Hovering over a segment shows you who’s in it
I’m looking for those providers whose names include any of these; exclude those
14. now we see a list of organizations we actually care about
This is a list stakeholders actually care about
You can track the growth of this audience segment over time
Think about applying this segment to your campaign tracking as well
Now let’s see what content they were looking at
15. Choose a secondary dimension.
In this case, if advanced segment is the WHO, dimension is the WHAT
16. Now we have the list of content this group of users was looking at over a period of time
Let’s scroll up the page
17. One of our stakeholders really cares about this data
Let’s automatically email them ONLY the data they care about
18. Every Monday, they see the report about which pieces of content some from HoR looked at in the past week
If you have trouble with content creators committing to producing new content, this helps
19. @devonvsmith
who sent them
to us?
Some visitors are self motivated - they had a specific topic or question they decided to search for
We’ll talk in a moment about how to increase your exposure to them
But other visitors are referred to you by some other organization, via a link
Part of your communications strategy should be how to influence those organizations, often by providing content to them
Let’s look at Referral Sources
20. Sometimes we see a big spike in traffic and it’s great!
Other times, it’s a spam attack
but in either case, this aggregate view of all traffic is like looking at an average - it buries the variation/outliers
Let’s add a few advanced segments to screen out some of that traffic
So I can pay attention to the links that actually matter
21. Let’s look at 3 different segments
I’ve pre-identified referral links from websites that are focused on job seekers, the media, or org partners
22. Now we can see the trends we care about
Easier to see unique trends for each of these groups
But the bottom section is still too messy for me, and includes a bunch of data I don’t care about
23. That’s better. I can see who the organizations are quickly.
For job seekers, I can start working with HR to focus on who’s sending most effective candidates
Start working with the PR department to ID which media sources we should be pitching stories to
Work with our programmatic team to know which of our partners share our audience & think content partnerships
24. @devonvsmith
What are they
looking for?
Let’s go back to that other half of visitors who are self-motivated to find a piece of information
If we can identify what they’re looking for, and provide it to them better, or increase our exposure to them
Organic Search Keywords
26. f
two problems: 80-95% (not provided) - which we’ll solve in a moment
& a long list of branded search terms
so let’s find a more interesting set of keyword queries your site visitors used
28. Now I have a list of questions users had when they arrived on site
Use this to adapt your content strategy
Write blog posts, create resources
29. @devonvsmith
What (else) are
they looking for?
But remember those (not provided) terms?
Google masks all organic keyword queries from users who are logged in to google products
You can see more of those terms if you’re an AdWords buyer
But this report will also show you
30. If you click this item and see this page, you need to take an extra step
33. Now you see a report that shows:
Impressions - how many times your site appeared for a user who had this query
Clicks - how many times they actually clicked on your site
Position - what position your site was in (Google is 10 sites/page)
Use this in an SEO strategy. High value keywords where your position or CTR is low
34. @devonvsmith
What are they
looking for
right [here]?
That was all about keyword queries visitors use in an external search engine to FIND your site
But once they get to your site, they often still can’t find what they’re looking for
Again, let’s focus on a single page
35. Typical to see 1-2% of visitors using internal site search
Let’s see what they’re searching for
36. Long list
Often, top of list is a [null] result because your search field is confusing
And jobs-related
For foundations, “logo” is common bc your grantees are looking for it
Let’s screen out those words we don’t need, so we can focus on what is important/different each month
38. now i can start to identify trends —> better promote content? new organization of content? new content?
content types (podcast, archive, data, documentary)
topics (arts, community engagement, voting)
people (fernando, alberto)
geographies (detroit)
39. Let’s hop back out to another report
Which pages are visitors on, when they’re searching for a particular term?
In this case, people aren’t finding what they need from “About”
40. Jobs are promoted elsewhere on the site
But these visitors are looking for jobs on the About page
Now we have a better understanding how to adapt the content strategy of that given page
41. For some of you, clicking on any report in this section looks like this - no data
Let’s fix this by going up to Admin
44. And now I need to find the parameter my internal site search is using to identify keywords
in most case, this will be q or s.
45. @devonvsmith
what do users need
from [this] page?
Secret: Most orgs, the majority of your users will not see your homepage during their visit
(BTW: use an advanced segment to prove that)
But your homepage is typically your most viewed page overall
So stakeholders care a lot about their place on the homepage
but what do your users actually need?
Let’s look at a Navigation Path
47. Google Analytics has an endless array of tabs that are hiding data from you
Navigation Summary shows you where users were, and where they went afterward
48. Let’s look at the top left corner -
when you can, turn percentages into people
3 in 4 people who see the homepage, are landing there as their first experience of the site
1 in 4 come from somewhere else in the site
For those, most come from about
But for now, I’m more interested in where they go TO next
1 in 2 leave the site entirely
For the others, mostly they go to see more About the organization
49. So how do we take a homepage like this, with dozens of links,
and simplify based on what users actually need
50. @devonvsmith
how much is
[that] action worth?
Goals - high hopes, low follow through
For the nonprofits, foundations, & gov I work with, a lack of goals implemented in GA is not your problem
A lack of clear goals in general might be
When you’re not an e-commerce, it’s harder to put a value on a goal
How much it’s worth to download a PDF depends on who is downloading
51. Even most of the largest orgs we’re working with don’t use goals, so you’re not alone
53. 4 types -
destination page (good for subscribing, donating, buying; or the last step in any multi-page process)
duration - how long someone spends on the site (good for orgs w/ ads; engagement models where you know the time it takes to
complete an action/lesson)
pages/session - just what it sounds like (good for apps, much harder for sites; not all pages are equal)
events - any action on the site that doesn’t generate a new URL (mostly use this one)
54. Destination page - end of an application process
Define a:
page,
monetary value (which you can make up, based on COMM budget),
multi-step process (to see where in the funnel users are dropping off)
57. Event
Define the event & its value
we’ll look at creating events in a few minutes, but these are 4 parameters applied to any given event
58. Goals Overview page, let’s look at the 4 goals this organization has implemented
59. Pretty typical list of high value engagement events for a think tank that cares about content
60. Once we’ve clicked on a single goal, we can follow it through any other report
In this case, I’m looking at social media sharing by device
Not surprising the conversion is much lower for mobile
But that has to do with the ease of sharing from a mobile device
62. For a mobile web app that had multiple steps a user progressed through
We identified a few places in the funnel where users were dropping out
Focused on A/B testing those pages
63. Better
data
@devonvsmith
5 pieces of data you can add to your Google Analytics implementation
I’ve included front end code in the first few to show as example
but it’s just a piece of the overall implementation (dependencies)
64. @devonvsmith
custom dimensions
In the past, we saw primary dimensions and secondary dimensions applied to our data
Custom Dimensions are new data points that you’re adding to Google Analytics
based on parameters that are defined in your CMS
Blog posts are a good example
Content drilldown organizes your site based on URL structure (slashes)
66. now we have a list of blog posts
But unless you have an editor for your blog, there’s probably no one in the organization that cares about every blog post.
in most cases, authors care about their own posts, or a program area cares about their posts
67. so let’s add “author” to this data (which is captured in the blog post, but not by default in GA)
68. now I see who some of my top authors are
and could send a report to them of only their blog posts
69. same idea, but now we’re looking at program areas (here called Category)
70. Now sorted by program area, I can easily export this data & build more relevant graphs
73. Defining is a simple decision about how frequently you want this data point
Every time it happens? only once per session? once per use? once per product?
74.
75. // Projects
var project = metaTag('project') || '(not set)';
return {
'hitType': 'pageview',
'page': customURL,
'dimension1': legacyID,
'dimension2': pageDate,
'dimension3': project,
'dimension4': engagementOptions,
'dimension5': wordCount,
'dimension6': imgCount,
'dimension7': mediaContact,
'dimension8': relatedTags
};
one example of capturing dimensions based on what ‘project’ it belongs to
what else you might think about capturing
76. @devonvsmith
event: Scroll depth
Webpages are much longer now
Concern about whether users are consuming the entire piece of content
Implement an event tracker that captures how far down the page a user scrolls
Again, event = no change in URL, so in this case, scrolling
77. for this org, we have 30 or so different events implemented across a dozen categories
Scroll depth is tracked for every page on the site
81. Now I see the pages that are viewed most often, click on a single page
82. Now I know what percent of users get what percent way down the page
Baseline - 40% don’t scroll at all (they never leave the baseline)
25% - 25% get a quarter down the page
100% - 5% make it all the way down to the last pixel of the footer
Let’s start setting thresholds for what “good” means (25% make it 75%, etc)
90. So let’s think about bumping that piece of content to the top of the page
Explains why so few people are scrolling to the footer - they don’t care (as much) about content at the bottom of the page
96. Here are our three different calls to action
Social is surprisingly low
97. But when we look back, social takes up far less real estate
If that’s a key metric, how do we make that action more prominent?
98. @devonvsmith
event: multimedia
one challenge for any analyst is pulling together data from multiple sources
for video embedded on your website, YouTube is always going to have the most comprehensive data, so it’s typically worth it to
log into their platform to pull data
But it won’t give you correlation about what else video watchers do on your site
100. now we see a list of videos that got any plays on the site
we’ve appended the video title and length of video to the event label for easy reference
using google tag manager
click on a particular video
101. Now we know roughly how far into the video they made it.
And start thinking about where the video sits in the page
The quality and length of the ideal videos on the site
105. Options for any data point in Google Analytics
But I typically skip straight to conditions
106. Define any single or set of conditions
So this segment would be only those sessions that included a PDF download
Can also set up ordered sequences: you came from social media, and then you shared a blog post
107. Pop back out to creating a segment
You can share this segment with anyone (we do for clients)
Or import from the gallery
108. but to be honest, i haven’t found very useful
110. Create a new filter to:
exclude your own internal traffic, vendor traffic
include only certain subdomains in certain views (segregate out a jobs subdomain)
111. Four types of filters
Also helps with cross domain tracking
112. Annotations
Annotations - leave breadcrumbs to yourself or other staff about what happened when
Let’s explain what happened on these two days since it so dramatically changed our traffic
114. No one has time to log into GA every day to see what’s going on
So give GA prompts to email or text you if something crazy like this happens
Go up to admin
115. custom alerts (aka intelligence events)
And set up a custom alert (GA also calls these intelligence events)
118. And this is the email I get now
Only bummer: next day (not real time)
119. shortcuts
We’ve adapted lots of reports this afternoon
But sometimes it’s a little cumbersome
Create shortcuts to reports that you find yourself building frequently
122. custom reports
Sometimes you can’t get a report to look exactly like what you want
Create a custom report that shows you only the data & graphs you care about
And add any number of dimensions you like
Where before we could only see blog category or author
123. Let’s set up a report to see both
I need to choose the metric group (essentially what you’re used to seeing on the right side of GA)
and a set of dimensions (what you see on the left side, and what happens when you click on a given one)
124. Now these are the only 3 pieces of data we’re looking at
Just blog posts, just page views,
128. what else can you do?
Content Groups - group related pages together
Cost Data Import - add marketing costs to calculate ROI
Experiments - A/B testing
In-Page Analytics - hotspots on the page
E-Commerce Tracking - product purchases
User IDs - track logged in users
Industry Benchmarking - see anonymized peer data
Session Timeouts - decide how long before a “log out”
Search Term Exclusions - permanently ignore branded
Channel Groupings - group related referrals together
Attribution Models - give values to your channels
Google Tag Manager - more FE code, easier deployment
Real Time - monitor events
language - group “en” together; c = bots
Seconds <10
130. Demographics and Interests data comes from the third-party DoubleClick cookie (for web traffic) and from anonymous identifiers
for mobile apps (i.e., Advertising ID for Android and IDFA for iOS).
Enable:
Enable Advertising Features for your property, and
Enable the Demographics and Interests reports for the view.
Analytics uses the same age, gender, and interests categories that you use in AdWords to target ads on the Google Display
Network.
131. Notice the % of total sessions in upper right
Don’t put an emphasis on these numbers yet
Key Metrics don’t offer “users”
132.
133. Affinity = lifestyle
In-Market = product purchase habits
Other = more detailed view (sub categories)
Any definition of an audience that you uncover in your Analytics reports can be turned into a Remarketing Audience that you can
use in AdWords.
144. Social sharing is another key action on the site
For this organization, particularly on the blog
But there were questions about where this social sharing functionality should be
So we set up a quick and dirty a/b test