With the ever-growing popularity of programmatic sales, ad ops and programmatic ops teams are faced with the enormous task of managing yield efficiently. With the seemingly endless amount of header bidders, ad exchanges, SSPs, RTBs, and ad network’s out there, all of which promise publishers “the world,” it takes a very savvy ad ops team to efficiently manage these various revenue streams. Ad serving technology has not yet caught up to the growing demands of the marketplace. The “low–tech” ad serving technology creates a lack of visibility for the publisher. With current ad serving technology, publishers cannot accurately create competition amongst partners. Publishers don’t always know their buyers or at what price inventory will be sold. So how do programmatic ops teams overcome this monumental task of managing yield with less than efficient ad serving technology?

Step 1: Reporting

A deep-dive analysis is necessary for each and every network partner, ad exchange, or SSP. The reporting should be broken up by site, web inventory, mobile inventory, mobile app inventory, video inventory, and any other pertinent means of inventory. With all the various demand sources out there, reporting can be a monumental task for any programmatic ops team. Ad serving discrepancies need to be monitored, and each demand source must be checked daily to ensure yield is being maximized.

In a nutshell, publishers should be focusing on the following metrics for each demand source:

  • Total ad requests (demand source and ad server)
  • Impressions filled
  • eCPM
  • rCPM
  • Fill rate %
  • Revenue
  • Response Time (header bidding only)

*rCPM is a statistic that many publishers do not calculate; however, it is one of the most important statistics in determining how well a demand source is performing. rCPM is a way to place each partner on a “level playing field” for analysis. Since it takes fill rate into account, this metric should be calculated manually using this formula:

  • CPM = Revenue/(TOTAL IMPRESSIONS/1,000)
  • Helpful tool for CPM calculation can be found here

In our example below, you can see that “demand source A” seems to be performing better than “demand source B”, due to the higher eCPM; however, the eCPM does not tell the entire story. When the rCPM is calculated, you can see that “demand source B” is actually performing at a better rate than “demand source A.” Therefore, “demand source B” is the stronger partner.

  • Example “Demand Source A”:
      • 10,000,000 impressions seen
      • 3,000,000 impressions filled
      • $1,000 of revenue
      • eCPM = $0.33
      • rCPM = $0.10
  • Example “Demand Source B”:
      • 5,000,000 impressions seen
      • 2,000,000 impressions filled
      • $1,000 of revenue
      • eCPM = $0.50
      • rCPM = $0.20 

Step 2: Ad Server Analysis

The programmatic landscape is getting more complex by the day, and ad serving technology simply cannot keep up. Therefore, it takes a very keen programmatic ops team to traffic these various demand sources appropriately. An overview analysis of the ad serving setup is a key factor in managing yield. The following settings must be noted and complied with the reporting metrics above:

  • Ad server priority/type
  • Ad server CPM/Rate
  • Ad server weight
  • Product catalog; formats and sizes
  • Start/end date
  • Frequency Capping
  • Daily impression caps
  • Geotargeting
  • Inventory accessed (desktop, mobile, in-app)

In order for a publisher’s programmatic ops team to efficiently monetize its ad tech stack, an ad server set-up in which all demand sources have a fair chance at a competition is recommended. In order to do this, for all tag based partners, a publisher must make priority and pricing adjustments based on a weekly basis to account for the ups and downs of the bidding activity. For more advanced publishers leveraging header bidding technology, a consistent setup between partners is recommended which includes price buckets and priority levels.

Step 3: Identify Top Performers and Under Performers

Not all demand sources are created equal. Therefore, when identifying performance, a savvy programmatic ops team will need to identify over/under achievers in the ad tech stack. It is recommended that programmatic ops teams review each partner’s reporting on a weekly basis to identify who is bringing value to the stack. Many times there will be a clear-cut reason behind a certain performance that can be controlled by the programmatic ops, and other times certain demand sources may simply not be a good fit for the publisher’s ad inventory. It is important to scrutinize each demand source’s reporting metrics in order to draw conclusions and make optimizations accordingly.

Step 4: Flattening the “Waterfall”

 After a deep dive into each demand source’s user interface, an investigation of your own ad server setup, and identifying your strongest revenue partners, the next step is to identify your “pass back waterfall” and floor prices.

In years past, publishers had no choice but to be caught in a never-ending waterfall of pass backs, and managing this was a huge task for ad operations teams. However, with the rise of programmatic, new technologies have emerged which allow publishers to “flatten” the waterfall. It is now possible to have all partners compete on the price the buyer is willing to pay, in real-time. Header bidding is a technology to avoid pass backs entirely.

Step 5: Implement Header Bidding Demand

Header bidding has changed the landscape. Header bidding is an advanced programmatic technique in which publishers offer inventory to multiple ad exchanges simultaneously before making calls to their ad servers (mainly DoubleClick for Publishers). The idea is that by letting multiple demand sources bid on the same inventory at the same time, publishers increase their yield and make more money. The rationale behind this strategy is to allow ADX dynamic allocation to compete at each price point, lifting overall programmatic revenue. Header bidding requires a savvy programmatic ops team, as well as savvy developers. It is considered a “hack” to the ad server and consists of pushing hundreds of line items to the publisher’s ad server at various price buckets. These line items only get “pinged” when there is a buyer for the inventory. Here is an example: if a buyer is willing to pay $5.50 for a user, that $5.50 price point gets pushed to the ad server and creates competition against other header bidders and Google ADX, driving up competition, increasing the publisher’s revenue.

Step 6: Create Competition

There is no “one size fits all” for a publisher’s monetization strategy. However, as a basic rule of thumb, in order to increase yield, a publisher must add quality competition to their ad tech stack. A publisher can accomplish this by implementing multiple header bidding demand partners, negotiating 100% fill flat CPMs with ad networks, and leveraging Google’s dynamic allocation feature to compete across the ad tech stack. Managing multiple header-bidding partners can create other issues for publishers; so many times a header-bidding wrapper is necessary. Managing header-bidding wrappers will be covered in our next post.

Step 7: Managing Floor Pricing

Historically, setting optimal price floors were the key strategy in establishing a solid ad tech stack. If a network could not meet the price floor, the next network would get a chance to serve the impression. However, waterfall setups have become less popular, and are not recommended since the advent of header bidding. A header bidder driven programmatic strategy changes programmatic ops in many ways. In a scenario in which every advertiser bids in the header, it is no longer necessary to create a waterfall. The highest paying advertiser gets the impression the first time. That means pass backs become irrelevant, and setting price floors becomes obsolete. If you are not yet leveraging header bidding, we recommend the following approach to setting floor pricing:

  • Ask your exchange partners to supply you with Bid Density Reporting.
  • This will give you the bidding behavior of DSPs on your site and will give you insight into what your inventory is worth on the open market.
  • Example 1: A top down approach to $3 floor on 300×600 inventory:
  • Drop the floor in $0.10 increments over a period of time (a week gives a good sample).
  • Ask for bid density reports on that unit.
  • As soon as new demand starts to wane, and you do not think it is worth the extra revenue, lock in the floor at this rate.
  • Example 2: A bottom up approach to no floor on open exchange 728×90:
  • Ask for a bid density report on this unit.
  • Isolate the $0.10 increment with the most bidding behavior.
  • Move floor to the pricing that doesn’t affect your bottom line but gives your inventory a premium in the market.
  • Stay aware of lower demand time periods and use top down approach to evaluate floors.


In summary, managing a publisher’s ad tech stack is more complicated than ever and requires savvy resources. Yet, as programmatic buying becomes more popular, there are new technologies emerging that are easing operations and allowing publishers to monetize their inventory more efficiently than ever before. With the right programmatic strategy, the sky is the limit. In short, it is an exciting time to be a publisher.