media-facebook/envy/_subsections/sec03/lesson-19.org
2025-08-15 15:46:31 +03:00

2.4 KiB

Section 03 - Lessons 19 | complete facebook split testing guide

Notes

split testing

  • test one variable at a time
  • A/B testing to see which ad performs better

    • Ad Set level
    • Ad level

manual split testing: ad set level

variables

  • interests
  • locations
  • age
  • audience
  • gender

how to

  • create the campain
  • duplicate ad set using button

    • you can now edit it (do one variable)

exclude audience

  • you must exclude the first campaign audience
  • otherwise you will have audience overlap (seeing it multiple times)
  • exclude audiences at the ad set level

    • exclude link under custom audiences and detailed targeting

example

  • two ads
  • two ad set audiences with a difference in some variable
  • therefore four ad sets

analyze campaign results via the breakdown menu

  • in the campaigns listing (ads manager)
  • on the far right you can choose variaous options to see who the ad appeals to

    • who clicked
    • who took action

coursenevy super split test

basics

  • enable Campaign Budget Optimization
  • each Ad Set requires minimum of $1 per
  • set daily budget to $30 to create 30 ad sets and test 30 different audiences

set up

  • create main campaign
  • navigate to ad set and hit duplicate
  • create 30 sets

    • don't worry about excluding for now

process

  • run it a few days
  • toggle off losers
  • scale up the budget on winners

suggested audeinces

  • anyone whose clicked on your CTA buttons in the past

split testing ads

  • duplicate the ad and test
  • dynamic creative

    • toggle at AD SET level
    • click on optimize
    • allows you to upload multiple assets and provide

      • images
      • videos
    • will mix and match and you can see who did the best

carousel reporting

  • you can see who got the most impressions in the breakdown

facebook ads 'A/B Test' Option

  • at campaign objective level you can have this done automaticall
  • toggle 'Create A/B test'

tips

  • test one variable
  • measurable hypopthesis (ie cost per action)
  • use big enough audience
  • set an ideal time frame
  • set an ideal budget