Top Header Bidding Partners Performance Report by Roxot
Sample & Methodology
The report is based on the Prebid Analytics by Roxot data from publishers who agreed to participate in the research. Prebid Analytics collects client-side data about all header bidding auctions running in the website's prebid.js wrapper and all demand partners participating. Therefore, the report is based only on the client-side data.
We analyzed almost 4 billion requests and 1.3 billion impressions generated by US users on 179 websites from July 15th to September 30th, 2017.
Please, mind that the report is based ONLY on the Prebid Analytics by Roxot users data and might be biased to some of the bidders. The users might have implemented demand partners incorrectly, which would affect the results of this report. However, we took measures to prevent any data inconsistency:
  • Reduced data contribution of the sites with total revenue share higher than 20%
  • Excluded sites with abnormal bidder behavior (e.g. 0 bids, bid rate lower than 5% or higher than 90%)
  • Excluded sites with less than 50% US traffic
  • Combined data of different adapters for the same demand provider (e.g PulsePoint and PulsePoint Lite, AppNexus and AppNexus AST, etc.)
Site Performance Analysis
Site Fill Rate* depending on the number of prebid.js bidders on a site
In the pursuit of maximized revenue, publishers try to increase the fill of their inventory by often blindly adding new partners. The research shows that, indeed, the Site Fill Rate correlates with the number of demand partners on the site. The biggest Fill Rate gap is between publishers who use 4-5 bidders and publishers who use 6-8 bidders.
*Site Fill Rate - how often there is at least one bid for an ad request
Site Impression Rate* depending on the number of prebid.js bidders used
The Site Impression Rate shows how prebid.js bidders compete with demand in an ad server (e.g. AdX or AdSense). In contrast to the Site Fill Rate, Impression Rate grows steadily with the increase of the number of bidders on a site. Although, adding more than 8 demand partners would not significantly increase your Desktop Impression Rate.
*Impression Rate - how often a Prebid.js winning ad is actually rendered. This metric shows how prebid.js competes with demand in the ad server.
Bidder Performance Analysis
Top 10 bidders by number of requests* received
The number of ad requests usually represents the level of bidder's adoption. The more publishers a demand partner works with, the more ad requests it receives. However, it might be a result of the bidder's simple integration process or loose integration requirements.
*Requests - a number of ad auctions a bidder was requested to participate in. Multiple requests for different ad sizes from the same ad unit count as one request.
Top bidders by Bid Rate*
The Bid Rate shows how often a demand partner replies with a bid to an ad request. If there is no bid, demand partner might be too slow or not have ads for a particular website visitor. In combination with the level of bidder's adoption, bidding frequency is a good indicator of demand partner's efficiency.
*Bid Rate - how often a bidder replies with a bid to an ad request
Top bidders by Win Rate*
The Win Rate shows how a bidder competes with other partners in the header. To get impressions and generate revenue a bidder has to both bid often and win often. A high Bid Rate would mean nothing if partner's bids never win.
*Win Rate - how often bidder's bids win prebid auctions
Top 10 bidders by the number of Impressions
The table below represents bidders' share of Total Impressions (ads that won both header bidding and ad server auctions). Comparing percentage of total impressions with the number of requests a bidder received, you can analyze a bidder's effectiveness. The bigger the difference, the less effective a bidder is.
Top bidders by revenue generated
The table below represents bidders' share of total revenue from rendered ads (Impressions) and shows how much each bidder contributes to publishers' total revenue. Some of the bidders might receive fewer requests, get fewer impressions, but generate more revenue than other bidders. Also, pay attention to the difference between desktop and mobile performance.
Conclusion
In collaboration with Prebid Analytics by Roxot users, we presented you a performance overview of the top header bidding demand partners. The data does not represent the whole market but it gives the clear understanding of the top bidders performance and implementation trends that can be used to adjust your monetization strategy:

  • The report shows that the Site Fill Rate (how often there is at least one bid for an ad request) correlates with the number of demand partners on the site. The biggest Fill Rate gap is between publishers who use 4-5 bidders and publishers who use 6-8 bidders (from 73% to 86% on Desktop, and 45% to 70% on Mobile). In contrast to the Site Fill Rate, the Impression Rate grows steadily with the increase of the number of bidders on a site.

  • Top 3 bidders by the number of requests received are Sovrn, AOL, and Index Exchange on both Desktop and Mobile. However, AOL and Sovrn demonstrate low Bid Rate and the inability to fill more than a half of requests. In turn, Rubicon and AppNexus have the highest Desktop and Mobile Bid Rate that demonstrates their high efficiency.

  • Index Exchange has the highest Desktop Win Rate (44.4%). It wins almost 30% of auctions it participated in (Bid Rate * Win Rate). Despite low Bid Rate on both Desktop and Mobile, Sovrn demonstrated 38.5% Desktop Win Rate which is the second highest result. However, Sovrn wins only 9% of Desktop auctions. Appnexus has the 37.3% Win Rate which results in 26% Desktop won auctions in combination with the Bid Rate.

  • As a result of the wide adoption, frequent bidding, and competitive clearing prices, Index Exchange and AppNexus are the demand partners with the biggest shares of total desktop impressions (16.3% and 11.7% respectively). Rubicon Project is the leader in the number of Mobile Impressions with the 25% share of all mobile Impressions.

  • AppNexus, Rubicon Project, and Index Exchange are the top revenue-generating bidders with 18.2%, 13.9%, and 13.9% share of total Desktop Revenue respectively. Facebook Audience Network and Rubicon Project are the leaders in Mobile Revenue generated with an astonishing combined share of the Total Mobile Revenue exceeding 55%.
About Roxot and Prebid Analytics by Roxot
Roxot provides the technology publishers need to make informed decisions and protect their position on the global ad market. Our technology solutions for data-driven optimization of your header bidding setup will help you quickly adapt to fluid market conditions and maximize website revenue.

Prebid Analytics by Roxot provides clean client-side data about all header bidding auctions running on the website and all demand partners participating. It allows you to monitor daily, hourly, and real-time changes in the header bidding performance, uncover issues with Prebid.js setup, and gain insights into increasing a website's revenue.
Getting started
Request a demo and one of our Publisher Relationship Managers will give you access to Prebid Analytics and walk you through the product.
Made on
Tilda