1: Testing Ads and Audiences (Walkthrough)

For context, I was aiming for $0.15 – $0.20 a click, since that’s realistic in urban fantasy and I was advertising a trilogy. In other genres or with more books, you may find higher click costs acceptable. To be honest, upon rewatching the video, I was surprised that I’d only done six tests. It felt like I tested way more things than I did: this process is frustrating and seems like a trek through the fog at least half the time, where every move you make seems pointless. Keep testing. That’s the only way you can either create good ads or determine that a particular book is not a good candidate to run ads to.

Finally, a key mistake: the testing process wound up costing me $200+ because I wasn’t punctual with coming in and making iterations. I could have gotten the same data for around $125. In the grand scheme of things, this isn’t a big deal; if these ads hold up during conversion, then they could run for months. But we don’t want to spend more money than necessary. This type of waste, when it crops up all over our business, can quickly turn into thousands of vaporized dollars.

Set a calendar date, set a timer, set an email reminder – do whatever you have to get the tests up and running in a timely manner. Part of the reason for the delays was because I had to record the video segments, and I couldn’t do so during the day due to external noise. But I want to emphasize that the faster you get through this process, the faster you’ll start seeing results.


  1. April 6: started with three ads, three ad sets
  2. April 8: let book background ad catch up, turned off one original ad, added iteration of best performing ad (headlines) [Test 1]
  3. April 11: turned off all ads besides book cover on book BG to allow it to catch up (was best performer), created two iterations on this book cover on book BG ad (copy), increased budget to $25/day to get data faster [Test 2]
  4. April 12: copied book background ad to completely new campaign, added four new audiences to test (audiences), new square image version of book cover on book BG [Test 3]
  5. April 14: iterated on book background ad and tried three stock photos (images) [Test 4]
  6. April 15: switched to multivariate testing instead of iteration, two new ads with new review quotes (headlines) and excerpts (copy) [Test 5]
  7. April 16: breakthrough to $0.15 CPCs, two additional stock photos (images) and three new audiences (audiences) [Test 6]
  8. April 18: ended up with three good ads and four or five usable audiences. Ready for next steps (aggregation, then conversion testing), covered in next videos.