About the client
A construction company from Belgium asked us for help in analyzing the sites of two residential complexes in Brussels.
Our Goal
The task was to split the traffic in half on two target sites. The client has two sites dedicated to their apartment complexes. Both sites have pages that correspond to equivalent customer databases. To preserve the client’s anonymity, let’s call them /1 and /2. These sections are identical to the user, but there is a difference in HTML, form tags and links.
The agency had a lot of traffic to the main page and almost no traffic to pages /1 and /2. So, the client wanted to split the main page traffic in half between two test pages. Depending on which page the user filled out the form, his data entered one of the two client databases, and this distribution should be even.
The customer tried to set this up using cookies in Google Tag Manager, but eventually turned to us for help.
Our Approach
First, we tried to redirect users to pages /1 and /2 based on GTM and a test in Google Optimize for the redirect.
Optimize is Google’s specialized tool for website and ad optimization. It helps test marketing hypotheses and test user reactions to various site changes.
So, we came up with a complex plan to split the traffic using Optimize and Google Tag Manager.
The essence of the approach was to divide the audience into two groups: new users and those who were already on the site. If we already have cookies, they contain a link to the page to which the user should be redirected (/1 or /2). Redirect should be done using javascript. If there were no cookies with a link, Optimize should have sent half of the users to one or another page through an experiment with redirection.
But it failed because the javascript redirect for old users conflicted with the rules of the experiment written in Optimize. During testing, we kept landing on the same page and it was unclear whether clearing cookies didn’t help, or whether Optimize was tossing a coin that kept falling heads.
Therefore, we used a simpler solution. Optimize has an A/B-test tool that can modify the HTML. And the client noted at the beginning that the landing pages do not differ from the user’s point of view, but have differences “under the hood”. Accordingly, we introduced all elements that differ /1 from /2 in the a/b test of Optimize.
While testing this solution, we learned that Optimize remembers the user not just by cookies — it is necessary to clear all temporary data, including the cache and switching of the Google account.
Result
Rethinking the problem, we presented our solution entirely based on Optimize, without third-party codes and even without Google Tag Manager. For two sites with three language versions each, we set up six experiment templates in Optimize and prepared a video for the client showing how it all works.
The client was satisfied with the result.
Conclusions
To create a simple analytical solution, the site owner does not need to learn programming languages and write scripts in GTM. The world is confidently moving towards the no-code trend, and this also applies to analytics.
This case was made on the basic version of Google Optimize, but in addition to it, there is also Optimize 360, a component of the Google Marketing Platform. It allows you to solve much more difficult tasks, and we, as certified partners of GMP, will gladly help you find these solutions.
