Today’s post is by Dana Kaplan, Lead Data Wrangler at Triggit.
When building out an effective retargeting campaign, choosing the right ad exchange and inventory is often a very important (and rarely discussed) factor in meeting your ROI goals. In addition to finding your intent-laden users out in the internet wilderness (Hint: FBX!), making sure your message is relevant (Hint: dynamic creative!), and all the other retargeting best practices (Hint: FBX playbook!), actually having enough opportunities from which to choose in order to serve your users ads is a key part of your performance. This begs the question: exactly how different is Facebook Exchange (FBX) from other exchanges when it comes to the volume of opportunities to buy impressions for each user?
To get an answer, we looked at a dataset of 1 million unique users, randomly sampled from our clients’ retargeting pools, and counted how many times they showed up for auctions in different exchanges for a single day. Our hypothesis was that because users browse Facebook pretty heavily when they visit and also tend to visit more than once a day — as opposed to general display ad inventory that could occupy a blog or news site article someone may visit only once a day — we will have much more biddable opportunities for each user we find on FBX.
In essence, if we find a user on FBX, we can expect to have more than 3.5x times the opportunities to bid on impression inventory for them compared to Google AdX. However, averages only provide so much insight, since there could be a subset of FB users who HEAVILY browse the site and show up for many more auctions; therefore inflating the FBX average.
To get a better sense of the volume of chances to bid on a user, let’s look at a frequency histogram of the dataset. The horizontal axis represents biddable opportunities per cookied user found on each exchange. The vertical axis is the % of each exchange that those cookied users make up:
Second observation: The percentage of cookied users found on Google AdX drops quickly, compared to FBX, as opportunities per user increases.
Looking at the chart, you’ll notice that in section “A”, which represents users who showed up for only 9 or less auctions per day, more than 60% of the cookied users found on Google’s exchange fall in that category. In sharp contrast, less than 20% of cookied users found on FBX appeared in 9 or less auctions. As we move into section “B”, the % curve of Google’s users drops suddenly and the FBX curve tends to spread out its % of users as auctions/user increases along the horizontal axis.
Illustrating the key takeaway from the above observations, here are the percent of users in each exchange who appear for auctions 10 or more times a day:
Essentially, when an advertiser finds one of their cookied users on FBX, they can expect to have MANY more biddable opportunities to show them an ad. What does more opportunities give us?
-The freedom to be more discerning how you bid and, in turn, more strategic and efficient buying per user:
-Play around with “frequency” and “recency” – perhaps you have found your campaign to be very effective when serving as many impressions as possible to your cookied site visitors in the first day of visiting your site, and then capping at a lower volume of impressions after that. Having more opportunities to bid affords us this flexibility.
-Be a patient bidder who waits for more efficient opportunities to retarget the right users while avoiding competition – Not only can you expect 3.5x more opportunities per day than other ad exchange inventory, but the other benefit of FBX is the consistent inventory ecosystem; being picky about price doesn’t mean the ad will be in any worse location or among worse content.
-As more advertisers enter the FBX channel, especially heading into Q4, we can expect competition won’t inflate CPMs as quickly as other channels:
-Remember, in retargeting, we are bidding for the user, not just straight impressions. If each user’s “supply” (biddable opportunities) is very high, then it can absorb more of the holiday “demand” (different advertisers bidding on that user) before prices inflate.
As Triggit’s “Data Wrangler,” it’s been interesting to look at how much better FBX has performed among more established channels like search and general display networks in standard metrics such as CTR, Conversion Rates, and CPC. Knowing that FBX is a different beast, with people browsing Facebook with a different mindset than other parts of the web, it’s important to try and see what behavior under the hood is generating this stellar performance. Being able to generate so many more auctions per user simply adds to the growing list of reasons why FBX is the most effective channel for direct response advertisers today.
-Dana Kaplan is Triggit’s Lead Data Wrangler and has worked in marketing & data analytics for more than 3 years. When he is not turning coffee into SQL, he can be found distance running, cycling, hiking, playing boardgames, and trying to impress people with his cooking.