by Doug

27/10/2017

Google Analytics

Scaled up testing using Google Optimise (Optimize)

Are you testing using Optimise 360? (disclosure, I’ll use the UK English spelling throughout)

I/We think it’s pretty bloody good.  From the setup interface to the performance, all very good.  The feature set for Optimise 360 is pretty compelling too.

So we use it. A. Lot. I mean so much we saw this recently:

Wait…What the?  Hang on, you may think, I can run as many tests as I like, right? No “credits” to limit my testing bandwidth. No arbitrary limits to hold back testing and p18n (personalisation).  Right?

Wrong.

Take a close look at the Optimise vs 360 documentation here: https://support.google.com/360suite/optimize/answer/7084762?hl=en

Up to 3 simultaneous experiments in the standard edition and More than 100* in Optimise 360.  What does the ‘*’ refer to?  Something you need to be aware of:

*Up to 24 experiments per Analytics view by default, more available upon request.

If you want to test at scale, get your request in via your reseller to get your limit boosted now!

There is a smarter way to scale testing efforts too.

Why so many experiments? We find there are use cases where you need to run completed tests at a 99%:1% level after a test has completed. You can observe and confirm long term winning treatment effects in this 99:1 mode. That can eat into your 24 quite quickly.

If you’re running a personalisation programme targeted at specific audiences (great!) then you can quickly find yourself in a position where you’ve got lots of p18n treatments in place but no more room for real tests without a boosted limit.

Here’s the fix: In your Optimise 360 account admin, look at the Container information section on the right hand side:

Yep, we’re deploying the Optimise tag via GTM in this instance – hence no completed step 4. Anyway, check your “Link to Google Analytics” section:

We’ve got two views linked. You may have only one.  Here’s the opportunity to scale your testing bandwidth.  The 24 simultaneous test limitation is per view. You can link more views for p18n and 99:1 tests leaving your main reporting view available for a full 24 real tests.

Hit the EDIT LINK (Why you shouting Optimise?) and add in a fresh new p18n view and a 99:1 view as you need. Then you got plenty of views to test against:

Worth mentioning also, is the view linking caveat:

Linking this container to Google Analytics allows data to flow between the two products. This may result in additional hits.

Watch out for your hit volume folks!

Now ramp up your testing – go learn and make the internet a better place.

Comments

Leave a comment

Your email address will not be published. Required fields are marked *