Can AdX Optimization Drain Money from Header Bidding?

The question is not surprising. Today’s programmatic trade is a jungle that mixes several demand sources and two auctions types. It’s absolutely not transparent nor is it efficient, so it’s not exactly easy to understand the optimization mechanics.

Yieldbird is a company that helps publishers increase programmatic revenue in many ways. Our main service is pricing management for Open Auction in Google Ad Exchange.

With the help of our algorithms and deep knowledge, we calculate and execute dynamic pricing strategies and tactics, which results in higher revenue for our clients. Of course, we work with publishers that also have header bidding and / or EBDA demand.

When we present our service for them, quite often they ask whether the uplift on Google AdX will be wasted by the drop on header bidding.

What Is AdX Optimization?

To understand the problem, we need to be on the same page on what we understand AdX optimization, or more generally, second-price auction optimization to be.

Do you like what you’re reading? Subscribe our newsletter for more content like this!

[mc4wp_form id=”407″]

Let’s take a look at all levels of floor prices in a standard auction.

• Neutral Floor Price: A price positioned under the second-highest offer. Neutral for cleaning price.
  • No Floor Price: this is just a special type of neutral floor price.

• Value Added Price: A price positioned between the first and second highest bid. In the second-price auction, you earn more from each auction if you are skilled enough or lucky enough to place the minimum price here.
  

• Blocking Floor Price: If you position your minimum price above the highest bid, you are not allowing the impression to be sold and you are losing some revenue.

Unfortunately, Yield Managers can’t set up a floor price after they get all the bids. They need to do it before the auction happens.

In addition, without data science and advanced engineering, it’s almost impossible to manage floor prices dynamically. This results in a situation where floor prices are not changed or regulated often. Sometimes they trade for days or even weeks.

As millions of auctions go through floor prices, each floor price generates some losses and adds some value. The same price is neutral, blocking, or value-added for different auctions.

We consider the yield optimization successful when the sum of blocked revenue is smaller than the sum of additional revenue generated by pricing.

Basic Header Bidding (or EBDA) and Google Ad Exchange Auction

Now, let’s add two more SSPs in the header to increase the complexity of the picture. Of course, we expect that new demand partners will bring us more bids per auction to increase auction pressure and simply let us earn more money.

Scenario 1: AdX is not optimized and the second bid comes from the header

Keep in mind that SSPs in the header are using first-price auctions, which makes things even more complex, but this is not changing the situation on the level of a single auction.

• On the header level, there is a competition between SSP 1 and SSP 2. The biggest offer comes from the second player.

• Then, Google AdX shows an ad and pays the “Header Competitor” Price + 0.01 cent. In this case, SSP 2 is pushing the AdX a bit.

Can AdX optimization generate a drop in header bidding or EBDA revenue?

No. In that case, Google AdX offers the highest offer and other SSPs will not win the impression. Let’s consider all pricing scenarios:

• Neutral Floor Price: The cleaning price will be not affected by the pricing.

• Value Added Price:
• If the floor price is lower than the highest header offer, it becomes neutral for the auction outcome.
• If the floor price is higher than the highest header offer, it is what we are looking for and we earn more money thanks to the pricing.

• Blocking Floor Price: Google AdX is blocked and the impression is sold to buyer “G” from the header partner. This situation has a negative impact on revenue.

Again, keep in mind that although all pricing strategies different than zero create all the above scenarios, effective yield optimizers are those that can ensure additional money generated by pricing is bigger than lost money.

To summarize this scenario:

• There is no way that header partners’ revenue can be reduced.

• When you apply any floor price for AdX (especially dynamic flooring strategies) and keep constant pricing on header bidding, you can expect additional growth on header-bidding revenue as well.

Scenario 2: AdX is (or is not) optimized and the highest bid comes from the header

This is easy. When one of the header buyers offers the highest bid, they win. Google AdX can be optimized or not, but as long as there is no magic, then no yield optimizers can charge a buyer more than they submitted to the DSP. As a publisher, you are paid exactly what the buyer bids.

Of course, also, in this case, header-bidding revenue can’t be affected by pricing management on Google Ad Exchange.

Scenario 3: Modifying the exclusive header-bidding range

When a publisher initially has the average floor price for Google AdX set higher than the average floor price for header partners, they create the space within which only header-bidding partners can win the impression. Google is simply blocked there.

Just after a publisher needs to modify Google pricing, they can increase or reduce the exclusive header-bidding range.

When Google prices increase, there is more space with no Google competition for header partners and they can spend more money.

Meanwhile Google pricing decreases, header partners face more competition. This is the only case when header-bidding revenue may drop, but as a publisher, you still earn more because all header bids that lose with Google lose because Google simply pays more for the impression.

Make sure

To make sure you earn more money, you can sum up all the revenue (like AdX and header) and compare them to some independent metrics, like the number of visits from Google Analytics.

One more thing: From the strategic point of view, the exclusive header-bidding range shall be reduced anyway. If you optimize AdX, you shall apply a quite consistent pricing for header demand. Buyers are smart and have much smarter tech than the sell-side.

If they see that it’s always cheaper to buy the impression through the header, they will buy through the header, reducing your revenue.

Who pays your bills?

Pricing is key to optimizing programmatic revenue, and good pricing always pushes advertisers to pay you more.

If you see more money after the pricing is applied to Google, you can be sure that it came from advertisers. Definitely, it is not wasted by the loss on the header side. Despite that, don’t forget the cost of optimizations.

No matter if you optimize with your existing yield team, hire a data scientist, or implement a technology that dynamically manages your pricing for Google AdX, there will always be a cost for optimization.

As a publisher, you should track additional revenue and optimization costs in order to keep them in a healthy relationship.

OK, this is a theory, but when I started AdX optimization, I saw header revenue drop

If you see something like this, then you probably have a technical bug. Especially if you implemented a technology that set up pricing dynamically.

Machines are much more efficient than people in pricing management. New implementations may be in conflict with existing header-bidding implementations.

The good thing is that all technical issues can be solved, so call your ad ops and developers.

Key Takeaways

• There are no auction mechanics in which AdX optimization can be wasted by the loss on header bidding. If you see something like that, there is probably a technical bug that your developers and adops can fix. However, AdX optimization itself produces more money by charging advertisers more, and thanks to AdX optimization, your total revenue is always higher.

• If the setup works properly, you can see less revenue from header bidding, but this drop is always compensated by the uplift on the AdX side. This situation may happen only when optimization requires reducing the exclusive header-bidding range, which for strategic purposes should be reduced anyway.

• Always track the cost of optimization. Make sure that it pays off to optimize your revenue and there is enough additional money to cover the cost of optimization and make your margins healthier.