To watch the webinar recording please find the link at the end of this post
Chrome’s announcement on the deprecation of third-party cookies has forced the industry to accelerate the efforts to build a new identification infrastructure that benefits both publishers and advertisers in the cookie-less future.
To prepare for 2021, Pubstack, in partnership with Webedia and ID5, has conducted an analysis of the impact of identity providers on publishers’ advertising revenues.
To prepare for cookies deprecation, Webedia integrated the Prebid Identity Module with ID5, The Trade Desk, and Criteo ID as ID providers.
To measure the identity module’s effect on their header bidding setup, they decided to use Pubstack to conduct A/B tests at scale through the Pubstack analytics dashboard and determine the incremental value of using an identity provider.
II- Methodology and Scope
Webedia’s Ad Stack
- Ad server: GAM;
- Header Bidding: Prebid and TAM with several partners connected;
- The test was conducted on 4 major websites: allocine.fr, 750g.fr, purepeople.fr, and jeuxvideos.fr, representing 40 M prebid auctions per day;
- The inventory was split into two populations; 50% were sent ad calls with an ID provider and the other without an ID provider;
- The test ran for 7 weeks;
- There was no discrimination by the ID provider.
This part will take you through the A/B test analysis results, whose goal was to compute identity providers’ predictive uplift on publishers’ advertising revenues.
PART 1: Assessing the impact of identity on Prebid Bidding
The goal here is to check if:
a- having an ID in the ad call allows Prebid to bid more often
b- the said bid has a higher CPM than when there is no ID in the ad call
To verify this, we’ve used the BPM metric, which calculates the value of the bids won by Prebid on the total number of auctions. This shows the real impact of identity without considering the competition outside Prebid (AdX, TAM, direct campaigns,…).
BPM by website
BPM = winning Prebid bid value*1000/auctions
From the graph above, we can see an uplift between 3,5% and 10% of BPM for the test population vs. the control population. This uplift can be due to two factors: an uplift in Prebid Bid Rate or an uplift in CPM.
Prebid Bid Rate by website
Bid Rate = auctions where Prebid answers with at least one bid/auction
CPM by website
As we can see from the graphs above, the CPM has no impact on the BPM uplift; it remains neutral whether there is an ID in the ad call or not.
PART 1 sum-up
To sum up this first part on Prebid bidding, we can conclude that:
- Identity seems to provide a 5% uplift on Prebid bidding;
- Identity makes Prebid bid more often in auctions;
- Identity seems to have no impact on Prebid bids’ value.
PART 2: Assessing the impact on Prebid Revenue
The goal here is to check if having an ID in the ad call allows Prebid to generate more ad revenues.
To verify this, we will use the RPM metric, which represents the prebid revenue for 1000 auctions. We will also use the CVR, which represents the ability of prebid to win an auction when confronted with the competition (ad server, tam, direct campaigns,…)
RPM by website
RPM= BPM X CVR
The graph above shows an uplift between 5% and 10% in prebid revenues for the test population vs. the control population. This uplift can be due to two factors: an uplift in CVR or an uplift in BPM.
After digging more in the data, we found out that there was only a slight uplift in CVR and that the uplift mainly comes from the BPM metric, which confirms further that Prebid bids more often when there is an ID in the ad call.
Identity has minimal impact on the ability of Prebid to win bids against the competition.
PART 2 sum-up
From the findings above, we can conclude that:
- The impact of identity on Prebid revenue is between +5% and +10%;
- Prebid doesn’t seem to have a clear advantage vs. the competition when it passes the bid thanks to an ID.
PART 3: Assessing the impact of identity by browser
We’ve pushed things a little further and wanted to know if there was a significant impact depending on the browser.
BPM and RPM by browser and by website
From the graphs above, we can see that:
- Chrome and Safari have the lowest uplift in prebid bidding (+2%), which is very surprising for Safari. This needs to be confirmed by further research.
- Firefox is the browser with the highest uplift in Prebid bidding (+10%);
PART 4: Assessing the impact of identity by bidder
We’ve tried to assess the impact of identity by bidder, but the results were not very conclusive. Probably because we couldn’t discriminate by ID provider, this hypothesis will be verified in future research.
If there were only four learnings to take from this survey, we would sum it up as follows:
- Identity providers bring between 5% to 10% in Prebid revenue uplift.
- This uplift mainly comes from additional bids and not from more valuable bids.
- Firefox is the browser that is the most impacted by identity, followed by Edge and Chrome/Safari.
- All main bidders are equally impacted by identity.
Ines Mouadil, Programmatic Operations Manager at Webedia, says:
“We are delighted to have worked with Pubstack on one of the first analyses of Prebid’s Identity Module. The results of this A/B test and the uplift in our programmatic revenues have confirmed the need to work with Identity in the programmatic chain. From a publisher’s perspective, we could never have measured this impact on our own without the help of Pubstack”.
If you want to have more details about the study, please watch here our webinar, where our Senior Data Analyst Adrien Rouget and Programmatic Operations Manager Inès Mouadil from Webedia are sharing the results. And if you are interested in understanding the impact on your own ad revenue please feel free to book a meeting with a Pubstack expert here.