Our First Facebook Canvas Campaigns: What We Learned
Published: April 19, 2016
Author: Mike Stetzer
Our team at 3Q Digital was one of the very first to test Facebook Canvas. We were given access and began building our Canvas test before it was available to the general public. That means we didn’t have blog posts or case studies we could reference; this was a brand new beast. Most people I’ve talked to or polled at the 3Q have still never seen one in their personal News Feed.
So how did we do with Facebook Canvas, and what did we learn? Read on to find out.
The client we ran the test with has a very strong foundation on Facebook. As a channel, Social drives the most conversions for them. Awareness is great, but CPA and sign-ups are ultimately how we define success. They’re also very cautious about trying new strategies; there’s always a mix of nervousness and excitement about results. “Hey! We’ve got a great thing going! Don’t screw it up!”
Because we had a great year last year, the focus this year is on how we can build on our success by trying new things (while also being cognizant of their goals). So for a change, they were VERY excited to test Facebook Canvas.
Ad testing on Facebook can be challenging because the platform and algorithm is much less mature than AdWords. We cannot, for example, select “rotate evenly” within the native Facebook UI and expect 3 ads to receive 33% of the impressions.
To minimize variables, we selected a Lookalike Audience that historically has performed well across a variety of Devices and Placements. That way, if performance on Canvas was underwhelming, we knew the audience was likely not to blame.
The nitty gritty: our Facebook Campaigns are broken out at the Placement and Device Level (ex – US Desktop News Feed 25+ or US Android Mobile News Feed_25+). In each of these campaigns, we have a 1% Sign Up Lookalike Audience, a 5% Sign Up Lookalike Audience, and a 10% Sign Up Lookalike Audience.
1% = Lowest Reach, but most similar to seed audience.
10% = Highest Reach, but least similar to seed audience.
We created a 3% Lookalike Audience and updated our exclusions, e.g. excluded 1% from the new 3% audiences; excluded 3% from the existing 5% audiences.
Given how important sign-ups and CPA are to our client, we had to keep other variables consistent in terms of optimizing for the same conversion pixel, driving to the same landing page, and utilizing similar bidding methodology.
- CTR (Link) on Canvas in March was 2.75%.
- CTR (Link) on Canvas was 587% higher than Account Average.
- Average user was spending approximately 25 seconds in the ad unit.
- Conversion Rate on Canvas in March was 0.75% — far lower than the account-wide CVR of 5.49%.
- CPA was approximately $15 more than our Evergreen campaigns.
- CPA was highest on iPad – it’s possible that the ad unit doesn’t render as well on these devices; however, CPA on iPad tends to be high for us in our Evergreen campaigns.
- CPCs (Link) were incredibly low at $0.46!
- Canvas CPC was 72% lower than mobile evergreen campaigns.
What We Learned
Facebook explicitly states on their website – “advertisers need a better way to share information after people click on their ad, and the information offered after someone clicks needs to load quickly, look beautiful on mobile and allow people to take action easily. We built Canvas – a new post-click, full-screen, immersive mobile ad experience, to solve this problem.”
The emphasis with Canvas is more on the experience vs. simply driving users to your website and asking them to purchase or sign up. When talking to clients in the future about Canvas, I’ve learned:
- To expect CVR and CPA (likely) to be worse than with traditional Evergreen ads.
- While Facebook calls Canvas “just another ad unit”, it’s still very new to users and shouldn’t simply be viewed as an additional layer to existing campaigns or ad sets. When testing, have Canvas live in its own campaign.
- Facebook’s website says that Canvas is “easy to create”. It’s not. We had a few designers and video production members on the client side who helped us build it.
- We highly recommend not taking existing assets and simply dumping them into Canvas. Take the time to build something clean and new.
- It’s important to pay attention to how Canvas might help boost performance in other channels, such as branded search or organic.
- We’re noticing in Google Analytics, % of new sessions is up month-over-month.
- In a way, Canvas is a nice blend between a traditional CTA-focused ad and a branding opportunity. CVRs will not be as good, but benefits can be seen in other areas.
So should you use Canvas? Maybe, but it’s important to have the right expectations: CPAs will go up, the units are hard to implement, but we did see a jump in site traffic.
Have you tested Canvas yet? Let us know your findings in the comments!