Tom Libertiny, Advertisement Testing with Artificial Intelligence

I met Derek Sivers a few years after he founded Since then, he sold the company and has lived all over the world, leaving us with great books and stories that he continues to write.

In this TEDx video, at the 1:30 time mark, he tells the story about product quality vs. quantity.

Summary:  Quantity can lead to quality.

(Continues after the video)

Advertising Testing

A/B testing (split testing) is when you run two advertisement versions promoting the same product or service.  The only difference between the two versions is one aspect.  It could be the”

  • Image
  • Video
  • Headline
  • Copy (text)
  • Call to action (CTA)
  • Voiceover
  • Music

Example A/B Test

By considering the first ad (A) the control and the second ad (B) the challenger, where the difference is the image, we ran both ads on the same platform and show them to the same quantity of potential customers for the same length of time.  In the most straightforward test, one ad would outperform the other by a measurement metric vital to you.

Examples include the number of people who:

  • View the landing page associated with each ad.
  • Purchase something that’s tracked through your website.
  • Sign up for your e-mail newsletter.

But how do we know we’ve chosen the best ad so we’re not leaving money on the table?

The answer is to run a series of A/B tests in parallel (a faster option, but not as accurate) or in-series (slower but more precise results).

Doing so means spending more time and money developing ads.

(Continues after the image)

Tom Libertiny, Advertisement Testing with Artificial Intelligence

Quantity Can Lead to Quality | Artificial Intelligence (AI)

AI can help by creating a series of variables to test in minutes.  Here’s that same list of variables that I usually test for and one AI solution for each of the items:

What I Did

Using these AIs, I created 30 advertisements, one control and the rest as challengers.

After three weeks, four advertisements outperformed the other ads by over 20% in revenue generated per ad view, the conversion rate (metric) we used to gauge performance.

With the assistance of AI, it took me one day to generate all 30 advertisements, which equates to less than 15 minutes per ad. Without AI, it would have taken me a week, including the time required to select appropriate stock images, video, and music, plus the licensing costs and the cost of finding and retaining a voiceover artist. 

Note:  The assumption here is that you have an ongoing business relationship with a number of voiceover artists and you receive priority in their work schdule and turnaround time.

Our fixed costs without AI would have been:

  • Images:  15 images (3 images per ad, of which three were for the control and 12 were for the challenger ads).  An off-the-shelf stock photo from Shutterstock costs $250 for a Standard License.
    • Image cost:  $3,750.
  • Video:  15 stock videos (3 videos per ad).  Shutterstock charges $80 per 15-second video for 1080p resolution with their Standard License.
    • Video cost:  $1,200.
  • Voiceover:  Top-rated talent through UpWork will be in the $100 to $150 per hour range.
    • Note 1:  It’s fair to say that a significant number of people view ads without the sound turned on (particularly true if people are primarily viewing your ad on a smartphone).  Including a voiceover for those who view your ad on a TV is still worthwhile.
    • Note 2:  We always include closed captions for those who have the sound turned off and for people with difficulty hearing.
    • Voiceover cost:  $300.
  • Music:  Background music can immediately set the tone for the ad
    • Note:  It’s fair to say that a significant number of people view ads without the sound turned on (particularly true if people are primarily viewing your ad on a smartphone).  It’s still worthwhile to include music for those viewing your ad on a TV.
    • Music cost:  $500.
  • Total fixed cost:  $5,750.

That’s an immediate cost savings of nearly $200 per advertisement.

Add to that the time savings for writing headlines, copy, and the CTA, depending on the fully loaded hourly rate for the people doing this work (in-house or externally), and the savings add up quickly when using Artificial Intelligence.

Note 1:  Cost of Artificial Intelligence monthly fees.  Suppose you’re producing advertisements at this rate. In that case, the monthly payments for the AIs I’ve listed are trivial on a per-ad basis.

Note 2:  If you currently use photographers and a video person or crew for visuals, your savings will be much higher.

Note 3:  Regardless of using stock material or hiring a team to create all the parts of an ad, your turnaround time will be much faster (a day) vs. multiple days or weeks from the start of your project until you can launch your ad campaign.  This time savings will allow you to iterate and refine your advertising faster.

Secrets to Success

  1. Artificial Intelligence can substantially reduce the cost and time to create a series of advertisements.
  2. Use a database, SQL, and AI to analyze your ad’s performance results in depth. Analytics tools, including Google Analytics, are helpful but not always accurate and require considerable knowledge and time to set up correctly.

Questions?  Contact me Today