First off, I highly recommend David’s work, his site, and his newsletter. The only issue I have with his work is style. He’s not exactly concise. He’ll go at great length to explain something and he makes generous use of the principle that saying the same thing in different ways helps more people understand. I’ve stopped reading some of his newsletters because of this, even though I know they contain some valuable tips.
I’ll note that David has a personal bias toward publishing exclusively through Amazon so sometimes he sees the world through that lens. He does try to be objective but admits that’s the core of his experience and expertise.
So I’m going to distill the newsletters I find interesting down into the very basics and post them here, more as a reference for my use than anything else. With that, I present my condensed version of two separate emails on the subject of targeting ads on Facebook, and a follow up on tracking performance:
- Campaign Goal: Traffic.
- Put only one ad set in a campaign. With different ads in the same set, machine learning will quickly start favouring one of them. Decisions are made too quickly, after just a few hundred impressions. 1000 impressions is more reliable.
- Set Advantage Campaign Budget off.
- Advantage Detail Targeting: off
- Interest targeting remains the best way to sell books at scale with Facebook Ads. Done right, they’ll remain the backbone of every campaign.
- Don’t consider Custom Audiences until you have experience and an audience of a few thousand readers on either their mailing list or Facebook Page (several thousand before attempting Lookalikes).
- Facebook defaults to the country you reside in. Switch to Americans, additionally those who shop at Kindle, then Kobo, etc. (unless going to your own site).
- Branch out, to separate ads and campaigns for each retailer and country (indeed, for each retailer in each country you are targeting).
- Make sure you are using the correct retailer link for the country you are targeting.
- Don’t hyper-target.
- Target one country at a time.
- Don’t target by age/sex/etc. unless you have a compelling reason.
Target by interest(s) only and keep the resulting audience as big as possible:
- Choose a comparable author or two, and if that’s not possible (it often isn’t on Facebook except for the biggest tradpub authors) then choose your genre.
- If that’s not possible you must get creative and start looking at things like TV shows your readers might watch.
- Narrow the audience by “Amazon Kindle” and the “Kindle Store” interests to ensure you are reaching Amazon customers who purchase e-books.
- Be careful with interests like “ebooks” and “ereaders” as they generally include lots of non-Amazon customers (see below).
- Don’t throw in too many interests at once – you need to test things individually first to see what works.
- Make sure interests are an intersection not a union by using the “Narrow Audience” button.
- Try all variations when searching for interests. “J.R. Smith” won’t match “JR Smith”
- Avoid general e-book interests: E-books (publications) and e-book readers (consumer electronics), not good for ads going to retailers, only to your site.
- There are two Amazon Kindle interest groups, one with 40M owners, the other with Kindle employees. Also Kindle Store with 25M members. Avoid Kindle Fire.
- Generally target audiences of sizes between 100,000 up to 5M-10M
- Use the Save This Audience button. Refine and then use “Save As” and create a new audience as you refine or target another country.
- DO NOT use Advantage Placements or Advantage Detail Targeting. Use interest suggestions instead and curate.
- Initially run ads on Facebook News Feed only, expand to just News Feed when more familiar with the whole process. Steer clear of Audience Network.
- Mix up formats in the same ad set (square 1080×1080, letterbox 1200×628).
Doesn’t recommend dynamic ads. Thinks the effort put into learning how to create a good ad is better.
Tracking and Adjusting
Ads have a life cycle. After some time performance will drop off. If you have multiple promotions across multiple platforms, tracking sales alone can’t tell you which part is starting to under perform.
You can track a few metrics:
- CPC. CPC can rise due to competition or because the audience is getting tapped out.
- CTR. An indication that people are becoming less responsive to your ad.
- Frequency (estimated). If frequency goes over 2 during a week, it’s a sign the ad is getting tired.
- Also useful: Delivery, Frequency, Budget, Amount Spent, Results, Reach, Impressions, Schedule, and Ends.
- Ignore all “performance suggestions”.
NB: The default metrics are misleading, track “Unique Outbound Clicks”, “Unique Outbound CTR”, and “Unique Outbound CPC” for a real measure of performance.
If the ad does appear to be approaching end of life:
- Consider simply pausing it.
- Duplicate the ad and try new targeting. Don’t change an ad that performs well because you will lose “social proof” (a history of good performance that FB considers when placing your ad).
- Duplicate the ad and try new media (images, copy, etc).
- Facebook will automatically A/B test two creatives in the same ad group, however it will bias heavily to the top performer. Consider disabling one if it starts to perform poorly. The decisions the algorithm makes are mostly good, but sometimes it makes that early and doesn’t revisit it later on.
Merge multiple well performing audiences into one and let FB do the targeting.
Authors Optimizing Amazon and Facebook Ads – Support Group (must have a published book to join) https://www.facebook.com/groups/393917614473395