Another quick post this week to show you how to set up an optimisation test with what seems to be Optimizely’s new competitor, Google Optimize (Part of Google Analytics 360 Suite).
The product is still in BETA, so not sure how many more additions they plan to make; however, at a first glance it seems that Google is ready to start getting some market share from the optimisation world.
Step 1: Setting up your Google Optimize Account
Before you start setting up a container with Google Optimize, you will need to download the Google Optimize PlugIn. Without it, you won’t be able to load the variant page for editing so make sure you download it from Google Store.
Once you install the plugin, you will need to create an Optimize container. Similar to Google Tag Manager, you will add your container’s name, the Google Analytics property you want to test on and finally the GA view you wish to see data on:
Once you do that, you will have to install the Google Optimizer snippet on your web site:
You will need to paste it into every web page you want to test. Paste it immediately before any other tags in <head> tag.
Alternatively, you can also implement the Optimize snippet via Google Tag Manager.
Go to your Google Tag Manager tag section and click on new tag, then select Google Optimize:
Then add your Google Analytics Tracking ID, the Optimise Container ID and select “all pages” as your firing trigger (see below). Save and then publish the tag:
Step 2: Setting up the Test
Once you have finished with step one, it is time to start setting up your test.
Google Optimize gives you three different testing options (as expected). You will have to click the blue icon right at the bottom of your interface.
You can run a Redirect Test, a multi variant test or an A/B test. For this example we will choose an A/B test:
Once you choose the test, you will need to give the experiment a name, then add the URL page you wish to test on the “Editor Page”.
Then select the Google Analytics property and the view for the experiment and click “Create”:
Step 3: Setting up your test objectives
Before moving on and editing your variant page, you will have a couple of choices in setting up your objectives (you can always design your variant and return to do that later). Google Optimize will scan your Goals and GA set up and recommend a couple of Objectives for you. For this test we will test the Bounces.
Click on the Primary Objective and select the metric you wish to run the test on. Of Course you have the option to set more than one objective if you wish (for example Bounces and Adsense Revenue):
Once you do that, move to the next option which is “Targeting”. The first option is to set up the percentage of visitors to target:
This controls how many people who visit your site see one of your test pages, including your original page. Everyone else will see your original page. If you want quicker results, you might want to include a higher proportion of your visitors in the experiment.
However, if your experiment is drastic and risky, you might want to include only a small proportion of your site’s visitors.
Now that you are done with traffic allocation it is time to create the testing rules.
Creating the testing rules
Google Optimize includes the following targeting rule types:
- URL targeting – Target the specific URLs where your experiments run.
- Audience targeting – Target Audiences created in Google Analytics.
- Behavior targeting – Target new vs. returning visitors or those coming from specific referrers.
- Geo targeting – Target visitors from a specific city, metro, region or county.
- Technology targeting – Target visitors using a specific device, browser or OS.
- JavaScript variable – Target a JavaScript variable in the source code of the webpage.
- First-party cookie – Target users that have a first-party cookie from your website.
- Custom JavaScript – Target pages based upon a value returned by custom JavaScript.
- Query Parameter – Target specific pages and sets of pages.
- Data Layer variable – Target based on key values stored in the data layer.
Note: If you go with “Audience Targeting” you will need to have Google Audience manager access. Currently I don’t, so I promise to come back later and show you how that particular audience type works.
Step 4: Designing your Variant Page
Now it is time to set up your variant page.
Similar to Optimizely interface, you will see your main page loaded and an option to create as many variants as you want.
In order to start designing your new variant, simply click on the +New Variant link:
Once you do that, select the device you wish to optimise for. This can be found on the top right corner of your editor:
For this example we will leave the test to Desktop.
Once this is set, your screen will look like this:
By clicking on the element you wish to edit, an editor will appear on the right of the screen where you can apply the changes you wish to test such as moving the location of the element, changing dimensions, uploading a new element on this position, run JS, change the destination URL, etc.
Funnily enough each time you click on an element a CSS selector will appear on the top of the screen. This is pretty cool as it can help you setting up rules in Adobe DTM assuming you are using it.
You can also edit the text/colour/position of each button or CTA.
All you have to do is select the item:
Then selection the option from the drop down menu:
Once you have saved your changes, your variant will appear to your main screen. You can now either launch the test or come back and edit it later (and of course delete it if you wish to start over):
Thoughts about the Google Optimizer (BETA) and whether it can make it in the Optimization world
This is a very interesting question to answer, so let’s get some facts straight first.
Optimizely has grown dramatically over the last few years (even though I am sure that this graph includes free accounts as well) with Adobe’s Test & Target following in second position:
In particular it holds 38 percent of the market share with Adobe Omniture’s Test & Target coming second with 19 percent base on latest buildwith data. However, we need to bear in mind that T&T is an enterprise solution and users can not use the tool for free, so I am not sure who is winning in terms of revenue.
Google’s old Optimization tool (website optimizer) held only one percent of the market share and I am not sure that anyone really bothered with its predecessor, Google Experiments. There were a couple of reasons Google Experiments failed, such as constant amends of the experiment code, lack of document support, no customer service and no ability to edit your tested page.
You had to create a new page to test on your own each time you wanted to launch an experiment. In short, Google Experiments would only allow you to do a test based on re-directing pages.
For SMBs who rely mostly to Google Analytics Free, this is going to be a good solution, assuming google will give a free account.
Update (01/10/2016): Google confirmed that next month they’ll start rolling out Google Optimize, a free version of the enterprise-class testing and personalization product, Google Optimize 360. You can request for an invite here.
For enterprises, this will come down to which analytics provider they use, since implementing technologies on large websites is way more complex and requires not only a know-how of different digital products but most importantly consistency in the technologies used especially if you are after a customer 360 view across all your real estates.
However there are cases where companies – despite the use of T &T – still go for Optimizely, due to the very good customer support they provide. Adobe is not known for doing great in that area and trust me this is something to massively consider when reviewing providers.
Having said that, I am not sure if Google plans to support Google Optimizer clients, or if you have to be part of a larger suit such as the 360 before you get support and account management.
From an implementation POV I don’t really see the difference between Google and Optimizely due to the use of the same library, so both tools tends to be compatible with GA. However,with some tweaking, implementation via Adobe DTM is also possible.