Case Study

EMR Bidding Algorithm Saves Participating Buyers More than $1M in Trial

May 21, 2018
By Tameka Kee, Director, Marketing & Content Strategy

In an industry known for non-stop change, the rapid rise of header bidding forced even the most adaptable companies out of their comfort zones. Nowhere has this been more apparent than in the shift to first-price auctions.

It’s been a jarring transition for many buyers, particularly since buying platforms and algorithms have been trained for years on second-price auction dynamics.

That’s why Rubicon Project developed Estimated Market Rate (or EMR), a feature designed to help buyers avoid overpaying for impressions in a first-price header bidding world. The algorithm aims to maintain high win rates while avoiding spikes in CPMs, and it can work in addition to (or in lieu of) a buying platform’s existing algorithm.

After months of internal prep and external communication, we moved to first price for all header-bidded inventory, offering buyers EMR to help navigate the transition.   

At the time of launch, just under one-third of the more than 150 DSPs we work with opted in. As of May 15, that number has risen to nearly half — and the results that EMR has delivered illustrate why.  

How we know it’s working

EMR looks for opportunities to reduce a buyer’s first-price bid with minimal impact to win rate. The algorithm analyzes factors like inventory type, target audience, and CPM, in order to come up with an adjustment to the buyer’s initial price.

Two key data points determine whether EMR is effective vs. the buyer’s initial bid:

  • Win rate increase: How many more auctions will the buyer ultimately win?  
  • CPM increase: How much more will the buyer pay per impression to secure that higher win rate?

Figuring out the sweet spot between how much a buyer should pay, and the probability of them winning the auction at that price is at the heart of EMR. Below, you’ll find two examples of how the feature has successfully delivered value for our buying partners.

Large, omnichannel DSP

This is what most people think of when they consider a “leading” buy side platform — a global DSP transacting tens of millions of impressions per day.  

The DSP buys inventory on behalf of Fortune 1000 brands and agency holding companies, seeking audiences across desktop, in-app, mobile web and video inventory.

How much traffic was allocated to EMR during the trial period?

  • Nearly 100% of the DSP’s traffic1 allocated to EMR

What were the results?

  • Roughly 1.8B impressions won
  • ~$1.02MM2 in efficiency savings (vs. what the advertisers buying through this DSP would have paid at first-price)

This trial highlights how Rubicon Project’s commitment to transparency fostered a strong sense of trust with one of our key buying partners — trust that ultimately led to a significant savings vs. a pure first-price strategy.

We gained that trust by working to educate them about how the shift to first-price might impact their business, which translated into a willingness to test our EMR algorithm for nearly all of the bids they made.

Emerging-format focused DSP

Our next example is a smaller DSP specializing in what could be considered an emerging format. This company targets audiences while they’re in lean back, relaxed mode, buying inventory on behalf of advertisers with gorgeous, interactive campaigns that warrant larger screens.

How much traffic was allocated to EMR during the trial period?

  • Roughly 25% of traffic allocated to EMR

What were the results?

  • Roughly 1.8B impressions won  
  • ~$3,500 in efficiency savings

This was a trial in a true sense of the word, with the DSP initially only opting-in to use the EMR algorithm for about a quarter of their buys.

As the weeks went on, the platform saw a massive difference in terms of the render rate for bids using EMR vs. their purely first-price bids. At the tail end of the trial, they elected to let EMR guide all of their buys.

The bottom line

We selected these examples to showcase EMR’s effectiveness and versatility.

It’s a feature that works for buyers with a variety of target audiences, formats and campaign goals — and it performs as well for large, sophisticated DSPs as it does for smaller, more specialized platforms.  

In the coming months, we’ll continue to evolve EMR to drive even greater performance for our buyers, helping to ensure that our sellers have a vibrant market of demand for their inventory.

Reach out to us to find out more about how EMR can help deliver value at contact@rubiconproject.com, or download our whitepaper for a deeper dive into how the algorithm works.


1 Rubicon Project reserves a small portion (~3%) of all buyer traffic for testing across multiple algorithms (including EMR, first-price, etc.) as part of our commitment to delivering advertiser value.

2 Savings calculated by multiplying the number of impressions served (from the auctions won using EMR) by the difference in CPM that would have been paid at first-price, divided by 1000.


Tags: , , ,