Facebook Automatic Placements: Considerations before Testing
Published: February 22, 2017
Author: Mike Stetzer
Throughout the last year or so, Facebook has been strongly recommending that we test Automatic Placements on our accounts (it even says ‘Recommended’ in the UI):
Facebook will automatically show ads to my audience in the places they’re likely to perform best!? The way that the description is written, it sure sounds like you need to try this approach. And as we recently heard from Facebook at their FinTech Summit, “For an individual ad set, we are always trying to serve the least expensive impression possible.” Facebook will generally favor mobile because it usually returns the cheapest impressions.
So: cheaper impressions will usually return a higher bang for your buck, and Automatic Placements will tend to favor mobile, as we discovered. But as the numbers below will show, we ended up having a very good reason to abandon Automatic Placements.
A single-client case against Automatic Placements
Before we dive into how and why we determined Automatic Placements should get anything but an automatic approval from us, a couple of caveats:
- The CVRs that this particular client is seeing on Mobile are (what I would consider) below average.
- Some clients have a fantastic UX on Mobile and in some cases, Mobile can actually convert better than Desktop. To that end, as the numbers will show, if those particular clients see traffic shift in favor of Mobile where they’re seeing better CVR and LTV, Automatic Placements might be great for them. In my experience, however, CVRs are generally better on Desktop.
Now for the numbers:
Before Automatic Placements
- Traffic Distribution
- 51% Desktop
- 49% Mobile
- CVR Comparison
- 6.8% Desktop
- 1.9% Mobile
After Automatic Placements
- Traffic Distribution
- 37% Desktop
- 62% Mobile
- CVR Comparison
- 8.2% Desktop
- 5.2% Mobile
Takeaways from the test
As you can see, Automatic Placements resulted in a pretty dramatic shift in traffic distribution towards Mobile (23% of total vs. 63% of total). This strategy was not sustainable for us; for this client, users are far more likely to convert on Desktop. We need to be spending where we’re seeing the best return for our clients — in this case, on Desktop. Separating Desktop and Mobile at the campaign level allows us to drive more Desktop traffic than we can with Automatic placements.
That said, with more and more users accessing Facebook via their phones/tablets vs. their PC, finding scale on desktop remains (and will continue to be) a challenge. This trend isn’t going away! The bulk of Facebook traffic is coming from mobile, and this will only get more pronounced over time. We are actively working with the client and our internal CRO Team to improve the client’s UX on Mobile.
There are other considerations outside of CVR — for example, Customer Lifetime Value (LTV). In many cases, clients see much better LTV from users who convert on Desktop. This can particularly be true for high-consideration products like loans or luxury retail. You should most definitely assess your LTV across devices before testing Automatic Placements. Note that even Facebook says they don’t recommend combining placements if your ultimate goal is LTV, because Facebook is only going to optimize for what they can see, which is the lowest cost per conversion within the platform.
What you can learn from our test
I suspect that with most clients/verticals (given the trends), Automatic Placements will result in a shift in traffic distribution towards Mobile.
Consider the following before testing this on your accounts:
- Is your Mobile UX and CVR poor?
- How do your Desktop CVRs compare to Mobile?
- How does your Customer LTV compare across devices?
- Do users who are served mobile impressions on Facebook have a high propensity to return and convert on Desktop? If so, Mobile may still be an efficient placement for your brand despite low CVRs. Pull a cross-device report in Ads Manager to determine this.
The upshot: don’t put blind trust in what you see in Ads Manager and Power Editor. Do tests as needed to determine what truly works for your brand.