When a Header bidding set-up has been successfully implemented, and all things are going “well”, a Publisher will often start to wonder what they can do to improve their revenue to the next level.
One question that we tend to receive in those situations is, “How do I manage my prices now, can I improve my revenue with a smart pricing strategy?”.
The question makes sense and different pricing strategies can have a real impact on revenue. But this is a question to consider cautiously because pricing management in a Header Bidding set-up is more complex than in a simple set-up. The key is to think “holistically”. There are now multiple interconnected platforms, and changing a price somewhere actually means changing the dynamics of the whole system… and possibly how buyers (or their algorithms…) behave versus your set-up.
Wait… prices in the header? What is this?
Managing prices is a “holistic” equation in Header Bidding and can have various impacts on your business. So the first thing to do is to start simple and clearly define the “pricing” landscape. There are multiple “price items” involved in a Header bidding set-up… Let’s go through each of these : these are the basic ingredients of a pricing strategy.
(Please note that this blog post specifically refers to Header Bidding client side. Server side header bidding tends to differ a bit and all hints and recommendations would not apply exactly the same)
- Minimum floor price in the primary SSP: the primary SSP is the SSP that is fully integrated with your Ad Server. The minimum floors there are based on the CPM that you implement in the pricing rules. If your Ad Server is DFP, then your primary SSP is Google AdX. As a reminder, let’s keep in mind that AdX is running 2nd price auctions.
In the Header, prices can be considered at different points in the “chain”:
- Minimum floor price in the Header partners: these are the pricing rules that you can define within each header platform.
- Floor in the wrapper: if you are using one, you can define floors in the wrapper. Meaning that bids from the Header partners below a certain CPM will not be taken into account and sent out to the Ad server.
- Floor in the Ad Server: the bid from a header partner is taken into account in the Ad Server and put in competition with the primary SSP only if it is targeted by a dedicated ad server line item (with a proper CPM). One way to “reject” header bids below a certain CPM is simply not to implement any line item in the Ad Server below a certain CPM.
Our initial recommendation would be to play with prices either at the partner, wrapper or ad server level based on what is the simplest for you from an operations standpoint.
Still, changing prices at one level rather than others will not have exactly the same effect:
- Hint #1: if your header bidding partner is running 2nd price auctions, then we recommend to manage prices at the partner level. The reason why is that “floor rules” at the partner level will have a direct impact on its auction dynamics. As an example, an increase of floor price can have a positive effect on CPMs due to increased pressure on buyers. Increasing the floor in the wrapper or the ad server would just reject low partner bids that are the result of a lack of competition and floor price within the partner auctions.
Also, depending on your set-up, managing prices at the partner level can give you more granularity: if you need to apply a specific floor CPM to a given brand industry, the only way to do it might be at the partner level.
- Hint #2: if your header bidding partner is running 1st price auctions, then changing the price at the partner / wrapper / ad server level has the same effect.
Still, at Adomik we always consider that more data is better! For that reason it might be better to manage prices in the Ad server so that no partner bid is rejected along the “chain” and all the bids make their way to the ad server logs. That might prove very useful in the future for analyses, troubleshooting and optimization purposes. Even if you are not involved in those types of analyses right now, the data remains available and can thus be used by yourself at any time. Still note that playing with prices in the Ad Server is also the solution that generates the most data “processing” activity, hence the most latency something that you want to be careful with if this is a sensitive issue.
- Hint #3: If you play with prices at multiple levels, be careful to properly align all prices at each level. Otherwise managing prices will likely be complicated and error prone, and might result in lost revenue.
How to evolve its prices?
All the ingredients are now known, let’s see how to use them. First, prices should not be used to optimize specifically one “platform”. In a header bidding set-up, prices have to be considered holistically. Otherwise they will affect the efficiency of the stack.
Here are 2 examples:
- If prices in the primary SSP are not managed consistently with those in the Header, this could cause Header/primary SSP competition issues (and as explained in the previous post, increasing competition is one of the key strengths of the header bidding setups). As an example if prices in the header are lower than in the primary SSP, then Header partners will win all impressions below the primary SPP floor price, without any competition. That will decrease the revenue of the primary SSP vs the header and might have a negative impact on overall revenue.
- If you change the prices for one partner in the header, you will then modify its behavior vs other partners. As an example, if you increase its floors, then you will lower the number of bids that the partner makes (it is only allowed to bid “high”, so it bids less in volume), and you will reallocate the revenue among all other partners and the primary SSP, making it complicated to understand the impact of the change.
The obvious conclusion to us is that when playing with prices, you have to comply with one basic rule: use the same floors for each header partner and within the primary SSP.
Having said that, while the rule is a good way to avoid making mistakes, there are some situations when breaking it might make sense:
- Hint #1: you might be willing to increase floors in your primary SSP. The usual rule would be to adjust floors in the similar manner in the Header. Still if “you know” that an increase of floor in the primary SSP will actually increase the revenue generated from that SSP and if you leave floors in the header unchanged (i.e. lower than in the primary SSP), you will increase as well the revenue of the header due to the lack of competition on low bids from the primary SSP. Hence a win win effect!
- Hint #2: if one of your Header partners is running 2nd price auctions, and is not generating enough revenue (potentially because it is bidding too low)… then increasing its floors is a good way to increase the price pressure and the competitiveness of this partner vs. others.
As you may notice, those 2 examples have to be taken cautiously for they can drive negative results if not managed properly. When you start to decorrelate prices in the header and your primary SSP, we strongly recommend that you have proper monitoring tools in place to measure the impact and adjust if necessary.
To be more precise, when we talk about monitoring the impact of a price change, you have to work on two different levels:
- After a price change, there will be an immediate effect of rebalancing revenue across your stack: between the header and your primary SSP (if not direct campaigns) and between header partners themselves. Make sure to understand how revenue, cpm and volume are moving around.
- Another impact will be how buyers react. In a header bidding set-up, changing prices can have an effect on how buyers behave, and decide to spend on one platform rather than others. If a buyer detects that one partner is cheaper than others then it might focus most, if not all, of its spending towards that partner. Call that a “cheap route”. To avoid cheap routes, a good rule of thumb is to avoid too big gaps between prices on the different partners and your primary SSP. To monitor such issues in your pricing strategy, the best is to have a holistic analytics product mapping all your buyers across your stack.
If you have any questions/comments, please reach out / request a product test!