Traffic is the basis for advertising monetization. How to make reasonable use of traffic and give full play to its maximum value is a problem that every advertising practitioner will face. This article discusses the distribution mechanism in traffic flow from the perspective of ADX. A reasonable distribution mechanism can maximize traffic benefits. I hope readers can get some inspiration from this article.
1. Traffic flow mechanism
ADX (AD Exchange), an advertising trading market, plays a connecting role in the traffic flow process. It connects upwards to DSPs and is responsible downwards to SSPs/media. Using its workflow to understand the advertising traffic flow mechanism will help us better understand the advertising traffic flow mechanism. Understand possible optimization points in the process of traffic overdistribution. The advertising traffic flow mechanism is as follows:
- When the front-end App triggers an advertising traffic opportunity, it will send this traffic to its connected ADX. The traffic attributes usually include related attributes such as advertising slots and user information;
- When ADX receives a traffic request, it will first verify the legality of the traffic. The simplest one is parameter verification, then verify the preset value of the order/DSP, and finally which DSP will deliver the traffic;
- When DSP receives this traffic, it decides whether to participate in the bidding based on the relevant attributes carried in the traffic. If the traffic is suitable, it returns the bidding price (or dealId) and advertising elements to ADX;
- ADX receives the bidding information from each DSP, and after a series of validity judgments, it is sorted according to the price bidding. The one with the highest bid gets it, and the winning advertising information is sent to the media, and the DSP is notified that its advertising has won (this step is not required) , but it is recommended);
- After receiving the advertising information, the media renders and displays the advertisement.
When user behavior occurs, ADX and DSP related behavior data need to be returned through the monitoring link, including main behavior exposure, clicks, downloads, wake-ups, etc.
There are two modes for returning behavioral data through monitoring links: C2S (Client to Server) and S2S (Server to Server). Currently, most customers require the C2S reporting method when launching.
The key indicators involved in ADX are mentioned in the ADX part of the previous article "Inventory of Commercial Advertising Roles". This article aims to explore the traffic distribution mechanism without explaining too much about the indicators. Interested readers can Read on the move.
Through the above traffic flow process, we can find that advertising traffic is mainly forwarded on the ADX side. If ADX is connected to multiple DSPs, a reasonable traffic distribution mechanism can increase the fill rate and ecpm, maximizing traffic revenue.
2.Waterfall
When ADX is connected to multiple DSPs, when requesting different DSPs, should the request be serial or parallel? There are different strategies involved here.
First let’s talk about serial requests, namely Waterfall. Waterfall, translated as "waterfall flow" in Chinese, literally means "flowing from top to bottom", but how should we understand the four words "from top to bottom"?
In the advertising industry, Waterfall refers to “requesting DSP from top to bottom to distribute traffic based on historical eCPM data when the value of each traffic cannot be evaluated in real time.” This is called an ad serial request.
Let’s look at the usage scenario of Waterfall through a practical example. Assuming that ADX is connected to three platforms, the eCPM and filler materials of the three platforms are as follows:
If there are 1,000 ad requests, there are the following ad request options:
Option 1: Request all DSP1
Profit = 1000 * 20 / 1000 * 30% = 6
Option 2: Request all DSP advertising sources
Profit = 1000 * 15 / 1000 * 50% = 7.5
Option 3: Request DSP3 for all
Profit = 1000 * 25 / 1000 *20% = 5
Judging from the above three solutions, although the eCPM of the solution is the lowest, its fill rate is the highest and the final total revenue is the highest. So is option 2 the best option? The answer is definitely no, because it only utilizes 50% of the traffic, and the remaining 50% of the traffic is wasted, so option 4 is derived.
Option 4: First request all 1000 advertising requests to DSP3, request the unfilled part to DSP1, and finally request the unfilled part to DSP2. The specific traffic distribution flow chart is as follows.
Profit = 1000 * 25 / 1000 * 20% + 800 * 20 / 1000 * 30% + 560 *15/ 1000 *50% = 14
Option 4 has a final income of 14 yuan and a fill rate of 72%. Compared with the first three options, it not only increases the income but also increases the fill rate.
Now that the revenue and fill rate have increased, can Waterfall be used to solve the traffic distribution problem? The reality will always slap you in the face. Waterfall’s solution mainly has the following problems:
- The core point of Waterfall is "historical eCPM data", so how to measure the historical eCPM data of a DSP? How long is this history?
- Serial requests will increase the ad display time, with an average request taking at least 100ms. Multiple requests will cause front-end display delays and poor user experience. Since the environment of different advertising slots is different and the user acceptance level is also different, the overall number of requests/timeout needs to be set for each advertising slot.
- Since Waterfall's request priority is determined based on historical eCPM data, for a specific request, the DSP bid at the front may not be as high as the bid at the back. In this way, you will miss the DSP ads with higher bids at the back, and the traffic benefits will not be maximized.
- For eCPM data maintenance of each DSP, due to seasonal issues, the eCPM value will change, which requires manual maintenance by operations students, which is costly.
Here, let’s talk about “eCPM of historical data”:
How long is this history? There is no standard answer to this question! Because the effect of each DSP is different.
The only thing we can do is to predict the eCPM and fill rate of each store. This can be verified through historical data or understood through business relationships. Only when we get correct and stable values can it be real and reliable for us. . 3 days, 7 days, 10 days or longer are all OK, as long as you think this number is reasonable and can withstand scrutiny.
For a newly connected DSP, since it has no historical data accumulation, how do we evaluate its eCPM value?
- You can learn about its eCPM and fill rate through business operation channels;
- Traffic support can be provided for the newly connected DSP, and after a certain amount of data is accumulated, it can be returned to the normal DSP for sorting. Each algorithm team has different requirements for this traffic support cycle and sample data, as long as it can meet their own business.
If two DSPs have the same eCPM and fill rate, how to sort them? At this time, you can evaluate it from other dimensions, such as interface response time, material quality, etc.
三、Header Bidding
Since Waterfall has many problems, are there any other alternatives?
Readers must be thinking that if the DSP can return the bid in real time every time a bid is placed, then there is no need to calculate and maintain "historical eCPM data". When distributing traffic, the traffic can be distributed in parallel. After getting After all DSPs place bids, the successful bidder will be determined based on the bids. This is "Header Bidding".
"Header Bidding", translated as "Header Bidding" in Chinese, literally means "the traffic is sent to the head buyers, the head media bids, and then uses the winning reserve price as the reserve price to request other DSPs that do not support real-time bidding." . To achieve this, the following prerequisites must first be met:
- When the top buyer returns the creative, he needs to return the bid at the same time, so that the media/ADX can complete the bidding;
- Although non-head buyers do not support real-time return bidding, they need to support the incoming advertising space floor price. In this way, if an advertisement is returned, the price must be higher than the floor price, which is the highest return for ADX and the media.
Header Bidding originated abroad and was first used on PCs.
DFP (Google Doubleclick For Publisher) is the most integrated advertising platform for foreign PC websites. Due to its monopoly on PC advertising and Google's Ad Exchange dynamitc bidding (for those who are interested, Baidu knows about it), it is very unfriendly to Publishers and other DSPs. Therefore, AppNexus hopes to join forces with other ADX/DSPs to shake the monopoly of DFP through Header Bidding technology.
From the above description, we can find that header bidding has the following advantages over Waterfall:
- Fair bidding: All DSPs bid at the same time and each evaluates the value of the traffic and bids;
- Maximize revenue: DSPs that were originally ranked at the bottom of Waterfall can win ad impression opportunities by increasing their bids.
In China, the development of PC has been relatively stagnant, and greater potential lies in the mobile terminal. Therefore, to be more precise, domestic header bidding should be called In-App bidding.
Due to the late start of domestic In-App bidding, currently only a few leading media support real-time return bidding. Therefore, header bidding and Waterfall will coexist for a long time. For media that support real-time bidding, priority will be given to header bidding. bidding, and then use the winning bid as the floor price of the advertising space to request other DSPs, and finally bid based on the price.
4. Summary
In fact, whether it is serial or parallel, they are just strategies to solve problems. The core goal is only "maximizing traffic revenue".
- From the perspective of the media, of course, we hope that more media will bid at the same time;
- From the perspective of a DSP, one must hope that the traffic will be sent to oneself first, and then the traffic will be sent to other companies after oneself has selected it, or even the traffic may be exclusive.
Of course, the real-life environment is complex, and different docking methods will also affect different strategies. Only by firmly grasping the focus of "maximizing traffic revenue" and taking into account the interests of multiple parties can we respond to changes without change.
- During the e-commerce festival, when major e-commerce companies compete for the market, their traffic budgets are sufficient. In order to get more budget, traffic is distributed to e-commerce DSPs first;
- The eCPM and fill rate of some DSPs are okay, but the materials are relatively low, and occasionally they may involve black-type ads, or there are technical pitfalls (such as high network latency). At this time, traffic needs to be targeted at these DSPs. limit;
- Although some DSPs have low eCPM, their fill rate is not bad, so they are more suitable for minimum filling and need to be supported by a certain proportion of traffic;
- ……
The author of this article @包子.
This article is submitted by an Internet user. The views expressed in the article represent only the author's own and do not represent the position of this site, and it does not assume any relevant legal liability. If reproduced, please indicate the source. If the content causes infringement/violation/factual inconsistency, please click "Content Report" below to submit a complaint!