This page provides a tutorial on the use of Google Analytics (abbreviated to "GA") to provide insight into selecting and validating the impact of changes to improve site speed and reduce Crashes and Exceptions.
Statistics on user abandonment, Bounce Rate and % Exit are also relevant because they indicate change in customer satistifaction and user productivity trends.
This work is different than the "performance" of AdWords and SEO techniques to yield customers.
Several efforts also improve page speed should yield improvement in user retention and revenue generated from the web property:
- Re-arrangement of UI elements
- Emails sent out
- Promotional offers
- Purchase of ads Google inserts onto websites
- Public news about spokesperson
Listed under the dashboard Behavior category, Google has an add-in only for its Chrome browser called In-Page Analytics at https://chrome.google.com/webstore/detail/page-analytics-by-google that displays stats on the very page being monitored.
PROTIP: Log out of your Gmail account before logging in if you use a corporate account.
Google Analytics (abbreviated to "GA" here) is at google.com/analytics.
The remainder of this tutorial assumes that you are logged in.
The product home for GA is at http://www.google.com/analytics
AppDynamics works by listening to what is occuring inside the data center. It can also push a RUM agent to end-user's browsers.
GA works by listening to data captured from end-users' browsers sent from from JavaScript snippets embedded in website HTML.
The JavaScript snippet can call a specific static tracking tag or call Google Tag Manager container code which dynamically sets when and which tags fire according to variables and trigger logic.
Tracking code can be asynchronous.
With Universal Analytics, Google takes a page load sample of 1% of page views, which can be increased.
GA reports do not rely on the User ID to which a user authenticates, but on a unique set of cookies stored on a particular person’s browser on a particular computer. So there are issues with cross-domain or subdomains as well as cross-device retention (when a user goes from desktop to mobile).
Note that GA may not be able to track all users. There is a growing movement of users who hate being tracked, and pushi back by blocking ads, trackers and GA. AdBlock Plus is the #1 downloaded Firefox add-on. Also, Ghostery blocks over 1,955 known ads, trackers, beacons and widgets, including GA.
Placing a badge/notice on your site announcing that you do not track users’ every move can attract users to come back and read/shop with you. (An edge case, I know, but still a positive one.)
-
http://analytics.blogspot.com/
-
https://www.youtube.com/watch?v=s4R5WWWg2Vk Google Analytics 2015 - Tutorial for Beginners nearly 2 hours from Lynda.com
Import the ** Site Performance Dashboard** by Justin Cutroni:
- Click the Reporting tab at the top of your dashboard home.
- Click on Dashboards on the left menu bar
- Click + New Dashboard
- Click Import from Gallery
- Click the search box at the upper right and enter Justin Cutroni.
- Click on Site Performance Dashboard (at time of this writing, it was last updated Sep 30, 2013).
- Click on the white Import button at the top of the page.
- Select the GA account.
- Click the blue Create button.
PROTIP: Metric values on their own are more valuable when compared against a different time range or other comparison.
1. Click on the date range at the upper-right corner.
2. For Date Range: select Last 30 days.
3. For Compare to: select previous period (Last 30 days).
4. Click Apply.
5. Click Save so the settings are retained.
PROTIP: View each number as a range of possible values. A certain number shown can be the result of pure chance, depending on the extent of variability such numbers have shown in the past. The more widely a metric varies, the less "reliable" a single number is.
PROTIP: GA does not provide 90th percentile numbers as AppD does.
Metrics most relevant to site speed, displayed on the Site Performance Dashboard include:
-
Total Sessions (count) PROTIP: When comparing stats, begin from the total number of Sessions.
-
Sessions (count) are the periods of time a user is active on a site (default 30 minutes).
ga(‘create’, ‘UA-XXXX-1’, ‘auto’, {‘siteSpeedSampleRate’: 50});
PROTIP: Mobile Page Load Time is usually higher than (total) Avg. Page Load Time because mobile networks (such a 3G and 4G) are generally (in 2015) slower than LAN/wi-fi/fiber speeds.
-
Mobile Page Load Time (Secs)
-
Avg. Page Load Time (Secs) is the time it takes for pages from the sample set to load, from initiation of the pageview (e.g. click on a page link) to load completion in the browser.
The number of resources a page loads vary based on screen design. So page-based metrics are not the best indicator.
ACTION: What can be done about network latancy?
- Design app
- Optimize size of transmission payloads
- Store data locally (but securely using Couchbase Mobile database to sync)
- Stage transaction data locally when there is no network
The ranking of how many page views by page name (such as "Home") is useful to know for load scripting to replicate load within a prod-like test environment so performance and capacity issues can be identified before ever affecting production customers.
-
Avg. Page Download Time is the time to download the page. This the mother of all speed metrics since it averages the amount of time it takes to completely load the page.
Page download is impacted by the speed of network used by the user and server response time.
-
Avg. Server Response Time is the time the server takes to respond to a user request, including the network time from user’s location to your server.
Additional metrics measure activities before pages load:
-
Avg. Redirection Time is the time spent in redirects before fetching this page. If there are no redirects, the value for this metric should be 0.
-
Avg. Domain Lookup Time** is the time spent in DNS lookup for this page. This should be <0.01
-
Avg. Server Connection Time is time spent in establishing TCP connection for this page.
-
Page Load Time by Browser
-
Load Time for Popular Pages
-
Avg. Server Response Time
Metrics most relevant to site speed include:
-
Active Users (count) have impact on the amount of memory consumed to maintain "state" of each user logged in the system.
-
New Users (count) have impact on the load on registration functionality which involves much database updates.
-
Avg. Session duration may be good or bad depending on whether tha app aims to keep users around (to view more ads) or aims to make it easier and faster to get work done.
-
Screens / Session speaks to the complexity of the application being monitored.
There is generally a correlation that longer sessions occur because of more screens being viewed by users.
This can be found elsewhere.
http://www.searchengineguide.com/jayson-demers/the-definitive-guide-to-google-analytics.php The Definitive Guide to Google Analytics for SEO Professionals
15 Google Analytics Tricks to Maximize Your Marketing Campaign http://www.forbes.com/sites/jaysondemers/2014/08/20/15-google-analytics-tricks-to-maximize-your-marketing-campaign/
The percentage of users who make it all the way to conversion "goal" concluding screen (such as receipt after buying, itinerary, back to Home, etc.) is a key influencer of customer satisfaction and website productivity.
How many users "gave up" because of too much complixity or too slow response.
First use of statistic may require some configuration, such as timeouts if average response times are high, such as:
1,038.88
## Crashes and Exceptions Crashes are a subset of Exceptions.To be actionable, crash statistics need to be segregated/filtered by application version and user:
- Under the menu category Behavior, select Crashes and Exceptions.
- Select Secondary dimension.
- Scroll down to Users.
- Select data metric User: Browser. Its value is always "Google Analytics".
- Select data metric User: Mobile Device Marketing Name.
The latest version may not be the one at the top of the list due to the pattern of app updates by users.
-
Select data metric User: Mobile Device Info. Mobile Device Marketing Name (such as "Galaxy S5") are a superset of the Mobile Device Info with values such as:
- Samsung SM-G900F Galaxy S5 (Samsung benchmark international model)
- Samsung SM-G900V Galaxy S5 (US carrier Verizon with non-Samsung firmware)
- Samsung SM-G900P Galaxy S5 (US carrier Sprint with Samsung firmware)
The difference between 41 individual models of just the Galaxy S5 are explained at http://androidforums.com/threads/41-galaxy-s5-models-dummies-guide.892162/
Some of these models have already had as many as 13 different stock Samsung firmware releases since launch.
-
Click the sorting arrow for data metric User: Mobile Device Info.
Notice the number associated with (Not Set). It's 6.11% of the total.
Time Category used:
-
Wait Time
-
Sign In is usually the most "expensive" (longest in duration) of all views even though it may not be among the most sampled (such as 17% of all User Timing Samples).
Each response time should be considered along with the rate processing when the metric was obtained, such as the number of transactions per minute (TPM), the number of transactions per second divided by 60.
Session statistics for users who share common attributes can be segmented. Segments include geographic region.
The number of visitors may include non-human bots.
GA provides output in various formats (CSV, PDF).
When showing each statistic, we want to compare statistics in GA against what work needs to be done For example:
-
when showing a list of mobile operating systems, compare the numbers GA calculates with the average response time for each operating system. Such information is used to allocate optimization effort across different dev teams.
-
when showing the percentage of New vs Returning Users, compare how many users were required to re-enter their credentials (statistics gathered from application logs). Such information is used to emulate load on authentication servers.
-
when showing a list of countries, compare GA's numbers with what allocations were targeted by marketing efforts, such as the percentage expected based on the targed population of each country. Such information can also be used to allocate use of remote CDN (Akamai) which places resources on servers close to where users are located.
GA provides a map overlay tab.
QUESTION: Why does both "en" and "en-us" appear?
Redirect Time for Countries
-
when showing a list of languages, compare GA's numbers with Such information is used to prioritize language translation efforts.
Some information, such as source/medium where traffic originated (such as another website or Google search engine). The medium, to the right of the slash, is the category, such as Facebook/Social.
To allows custom programs to make HTTP requests to send raw user interaction (event / hit) data directly to Google Analytics servers, Google provides its GAMP API at https://developers.google.com/analytics/devguides/collection/protocol/v1/
But what about monthly extracts for presentations to management and colleagues?
- Select the Month time scale. Notice the total before filtering.
Be aware of daily, weekly, and monthly cycles of normal activities. The patterns may be disrupted by special events such as the Superbowl, Thanksgiving, Christmas, etc.
http://www.analyticscanvas.com/ is a Windows-based client program from Google Analytics Partner nModal solutions in Toronto was highlighted in 2011 at http://analytics.blogspot.com/2011/11/simplifying-ecommerce-reporting-across.html