SlideShare a Scribd company logo

Create A/B tests with the Google Tag Manager

•Download as PPTX, PDF•
•
Jules Stuifbergen
Jules Stuifbergen

This document discusses how to set up A/B tests using Google Tag Manager. It explains two ground rules for A/B testing: 1) visitors are randomly assigned to variations, and 2) visitors stay in the same condition. Commercial tools like Optimizely and VWO are mentioned as doing fancy GUIs and ugly code well. The document then discusses replicating A/B test logic in GTM by assigning visitors randomly to conditions A or B, storing this in a cookie, and loading the proper code variation. It provides details on configuring the test, assigning visitors, using cookies, and triggering events and analytics. The goal is to allow setting up A/B tests for free using GTM's functionality.

Related slideshows

SEO proposal- www.thecorporateservices.com
SEO proposal- www.thecorporateservices.com SEO proposal- www.thecorporateservices.com
SEO proposal- www.thecorporateservices.com
 
Smo(social media optimization) ppt
Smo(social media optimization) pptSmo(social media optimization) ppt
Smo(social media optimization) ppt
 
SEO Audit Report Format
SEO Audit Report Format SEO Audit Report Format
SEO Audit Report Format
 
1 of 27
SET UP (AND ANALYZE) YOUR A/B TESTS LIKE A BOSS USING THE GOOGLE TAG MANAGER JULES STUIFBERGEN – CONVERSION HOTEL 2014
TOPICS FOR THIS SESSION • Two ground rules • The added value of commercial tools • What type of A/B tests am I talking about • Can we replicate this functionality (using GTM) • For free • For kicks • Because we can! • Let’s find out!
A/B TESTING RULE 1: RANDOMNESS Visitors enter your experimentation zone They’re assigned to a variation (A or B) randomly. Then, they’re part of your experiment.
A/B TESTING RULE 2: CONDITIONS STAY THE SAME Visitors are in the same condition, always. Condition A Condition B
THE ADDED VALUE OF COMMERCIAL TOOLS Optimizely VWO
THESE TOOLS DO 2 THINGS REALLY WELL UGLY code Fancy GUI
WHAT TYPE OF TESTS AM I TALKING ABOUT • Same URL / page group • Content manipulation with Javascript / jQuery Meet jQuery. A big fat Javascript library
LET’S REPLICATE THE LOGIC Pseudo code for a visitor entering the experiment 1. Does the visitor have a valid cookie? 1. Yes: set the condition variable w/ cookie value 2. No: Roll a dice and assign condition A or B 2. Store the condition in the cookie 3. Load the code for A or B 1. Javascript / jQuery code 2. Google Analytics code
READY TO CODE YOUR TAG?
DETAILS: CONFIG This code allows endless amount of variations. One cookie per test
DETAILS: ASSIGNMENT In words: If the cookie value is not in the list of the variation names: Pick a random variation.
DETAILS: COOKIE Set the cookie, always. This makes sure the cookie is renewed for 90 days (code stolen from the interweb)
DETAILS: THE FUN STUFF This is where the GTM magic begins. abtest macros for category, action and label (from datalayer)
DETAILS: THE FUN STUFF This is where the GTM magic begins. Built-in event macro, to trigger the action
RULE NR. 1 Fires on all abtest events
MEASURE LIKE A BOSS
RULE NR. 2 (AND 3..) Fires variation code
CODE LIKE A BOSS
OR.. CHEAT LIKE A BOSS Steal (and clean) optimizely’s code using the “edit code” functionality
START YOUR EXPERIMENT This goes in a Tag Activated by a Rule
Create A/B tests with the Google Tag Manager
Create A/B tests with the Google Tag Manager
DEBUG YOUR CONTAINER VERSION LIKE A BOSS
WELCOME VISITORS.. http://datadatadata.nl/ch2014.html
AND ANALYZE THE RESULTS Use Google Analytics User Segments on Events
YOU WIN!
THANK YOU Jules Stuifbergen @zjuul

More Related Content

Create A/B tests with the Google Tag Manager

  • 1. SET UP (AND ANALYZE) YOUR A/B TESTS LIKE A BOSS USING THE GOOGLE TAG MANAGER JULES STUIFBERGEN – CONVERSION HOTEL 2014
  • 2. TOPICS FOR THIS SESSION • Two ground rules • The added value of commercial tools • What type of A/B tests am I talking about • Can we replicate this functionality (using GTM) • For free • For kicks • Because we can! • Let’s find out!
  • 3. A/B TESTING RULE 1: RANDOMNESS Visitors enter your experimentation zone They’re assigned to a variation (A or B) randomly. Then, they’re part of your experiment.
  • 4. A/B TESTING RULE 2: CONDITIONS STAY THE SAME Visitors are in the same condition, always. Condition A Condition B
  • 5. THE ADDED VALUE OF COMMERCIAL TOOLS Optimizely VWO
  • 6. THESE TOOLS DO 2 THINGS REALLY WELL UGLY code Fancy GUI
  • 7. WHAT TYPE OF TESTS AM I TALKING ABOUT • Same URL / page group • Content manipulation with Javascript / jQuery Meet jQuery. A big fat Javascript library
  • 8. LET’S REPLICATE THE LOGIC Pseudo code for a visitor entering the experiment 1. Does the visitor have a valid cookie? 1. Yes: set the condition variable w/ cookie value 2. No: Roll a dice and assign condition A or B 2. Store the condition in the cookie 3. Load the code for A or B 1. Javascript / jQuery code 2. Google Analytics code
  • 9. READY TO CODE YOUR TAG?
  • 10. DETAILS: CONFIG This code allows endless amount of variations. One cookie per test
  • 11. DETAILS: ASSIGNMENT In words: If the cookie value is not in the list of the variation names: Pick a random variation.
  • 12. DETAILS: COOKIE Set the cookie, always. This makes sure the cookie is renewed for 90 days (code stolen from the interweb)
  • 13. DETAILS: THE FUN STUFF This is where the GTM magic begins. abtest macros for category, action and label (from datalayer)
  • 14. DETAILS: THE FUN STUFF This is where the GTM magic begins. Built-in event macro, to trigger the action
  • 15. RULE NR. 1 Fires on all abtest events
  • 17. RULE NR. 2 (AND 3..) Fires variation code
  • 18. CODE LIKE A BOSS
  • 19. OR.. CHEAT LIKE A BOSS Steal (and clean) optimizely’s code using the “edit code” functionality
  • 20. START YOUR EXPERIMENT This goes in a Tag Activated by a Rule
  • 23. DEBUG YOUR CONTAINER VERSION LIKE A BOSS
  • 25. AND ANALYZE THE RESULTS Use Google Analytics User Segments on Events
  • 27. THANK YOU Jules Stuifbergen @zjuul