The Ultimate Guide to Using Google Analytics for Cross Device Optimisation
Part 5: Google Analytics Templates
This is Part 5 of a 7 part guide. Parts 6 & 7 are coming soon. All the articles can be found in the Article Index
If you’ve read the main articles, great! This is just a nice roundup of all the reports and custom reports you can use to get the figures.
If you just arrived here, this article can’t teach you what to watch out for when analysing the reports below. If you don’t understand what data to trust or disregard — you might miss something.
If you know what you’re doing, you’re just here to get the report links and will never read this sentence, so I can spend more time takin’ your mommah clubbin, oh yeah.
The Diagram
Here are the reports to grab each datapoint to make your own diagram. I’ve made these reports very basic for a reason — because you want to be careful mixing pages and sessions in a custom report. If you want to know more, read here:
http://help.analyticsedge.com/googleanalytics/misunderstood-metrics-sessions-for-pages/
For now, I’m going to give you User Counts and Unique Pageview Counts in these report templates.Why not sessions?
Because of that article above that you didn’t read <grin>
Sessions can’t be used with the page dimension because it undercounts the number of sessions in a report like this.
Unique Pageviews actually turn out to be a very good proxy for sessions instead, because they count whether that pageview turned up within a session, one or more times!
Desktop
- Desktop:Windows
- Desktop:Apple
- Desktop:Linux
- Desktop:ChromeOS
https://www.google.com/analytics/web/template?uid=yDlq9WCiTBmwqr_bAobxxg
Make a note of the user counts (if you’re making a user model) or UPV counts (if you’re making a session model).
Desktop Segments
To get these, use the same report as above:
https://www.google.com/analytics/web/template?uid=yDlq9WCiTBmwqr_bAobxxg
And drill down into each operating system type to get:
- Desktop:Windows:Chrome
- Desktop:Windows:Internet Explorer
- Desktop:Windows:Firefox
- Desktop:Windows:Edge
- Desktop:Windows:Opera
- Desktop:Apple:Safari
- Desktop:Apple:Chrome
- Desktop:Apple :Firefox
- Desktop:Windows:Opera
Repeat the exercise with Desktop:Linux, Desktop:ChromeOS or any other splits with decent traffic. Don’t sweat the tiny percentages.
Desktop Subsegments
If you want to know the top versions for all those desktop browsers, you will have to read the article for how to do this:
Data Crunching Desktop Browsers
Tablet
- Tablet:Apple
- Tablet:Droid
- Tablet:Windows
Use this report to get these figures:
https://www.google.com/analytics/web/template?uid=yDlq9WCiTBmwqr_bAobxxg
“iOS” devices are the Apple tablets. “Android” is your droid figure and “Windows” is the count for Windows surface tablets.
Make a note of the users and UPV counts
Tablet Segments
- Tablet:Apple:[iPad model]
Sadly this data just doesn’t exist for Apple devices. If you want to figure out what models of iPads you have on your site, you’re stuffed (AFAIK, Feb 2016).
- Tablet:Droid:[Droid model]
If you want to get the droid models, you’ll have to read this article to get splits:
Tablet:Windows:[Windows model]
Windows models you’re stuffed too — as they all seem to record as “Windows RT tablet” — without any distinguishing info to show the model. I’d be interested to see if you pick up any other signals here you can share with me!
Mobile
- Mobile:Apple
- Mobile:Android
- Mobile:Windows
- Mobile:Blackberry
Just pull these figures using this report configuration:
https://www.google.com/analytics/web/template?uid=TUeBTcgXT7aOiXSyTZHDVg
“iOS” devices are the Apple phones. “Android” is your droid figure and “Windows” is for Microsoft (typically Nokia Lumia) handsets. “Windows Phone” is the older type of Windows OS. Most of the rest are too small (<1%) in key markets to bother with.
Mobile Segments
- Mobile:Apple:[Model name]
DO NOT use the Mobile device info or Mobile model name to get this value — until you’ve verified how many phones are identified as generic ‘iPhone’.
As of Feb 2016, my data looks wrong and most phones are ‘missing model information’. Use the screen resolution as a proxy — it’s a far better indicator of phone model/screen size.
- 320 x 480 = iPhone 4/4S
- 320 x 568 = iPhone 5/5S
- 375 x 667 = iPhone 6
- 414 x 736 = iPhone 6+
Use the report above and set the secondary dimension to ‘Screen Resolution’.
Enter the word ‘iOS’ in the filter box on the right hand of the report and hit [RETURN]. For more info on apple model identification, visit this article:
Mobile Device Model Data Crunch
- Mobile:Android:[Model name]
Clustering android devices by model name is misleading — because you just have zillions of manufacturers, but no way of grouping similar(ish) phone capabilities — like screen size & density, resolution etc.
There’s no easy way to cluster around, say, the resolution dimension in Google Analytics, because it’s broken. If you want to find the clusters in your Android models, which are of value to your work, read this:
Mobile Device Model Data Crunch
- Mobile:Windows:[Model name]
This doesn’t seem to be much help either — in GA — as the Mobile Device Info and Mobile Device Model dimensions seem mainly to be “(not set)”. Meh.
If you set the screen resolution to be the secondary dimension, on the report config above, you will see that we can at least use that as a rough guide for any testing we do.
If you’ve got a large Windows audience, at least you can try phones that have similar resolutions.
Download the Template!
Hope this is a useful short version of the full articles and that the reports helped get you the data to make this diagram.
Here is the Powerpoint template for the diagram (dropbox link):
You should make two versions — this first one has the relative splits of the traffic:
And this one, showing the absolute percentages. That middle column with the splits is how you should focus your x-device optimisation:
In this case, you could completely shaft this company by pursuing a cruddy mobile first strategy or some similarly rubbish technique guaranteed to ignore that 70% of visitors here are from desktop.
It’s a multi device world — get used to it and get some data, so you don’t kill the golden goose.
Optimising the Mix
Now comes the fun bit — we’re going to use our rich understanding of all the strange ways the data model is borked or fuzzy — and pull reports that will let us find the stragglers.
In all your sites, these lousy device experiences are happening. If only you could pluck out the bad performers and inspect them eh?
Knowing what traffic I have is one thing but how much money do I make (or lose) on these device experiences. So that’s what we’ll do in the next article — narrow the target list to where the delight or money is most likely to be found.
Coming soon — Part 6 — “Finding the Missing Money”
