| Tuesday May 24th 2016

OpenX: More Than Just an Ad Server

Ad-based commercial media networks are always busy with the constant improvement and fine-tuning of the network’s ad serving. It is essential to get the most dollars out of every impression that the network generates. OpenXUsually, someone is hired to improve the performance of the ad serving technology. Let’s call this person “The Optimizer”.

The optimizer’s job includes many aspects. His goal is to increase revenue. He needs to decide how to allocate the inventory to the variety of ad networks that deliver ads. Sometimes it’s necessary to give priority to a certain ad network or to change the placement of a non-performing banner. It’s a dynamic daily job that demands a close and accurate understanding of the metrics that tell the story.

The critical decisions made by the optimizer must be based on facts, which come in the form of metrics. Metrics reflect the efficiency of advertising over the network. Important metrics are CTR (Click Through Rate), RPM/CPM (Revenue Per Mille/Cost Per Mille) and Fill Rate.

  • CTR is the ratio of the number of clicks to the number of impressions per unit – the higher the better. One way to achieve a high CTR is to deliver ads relevant to the content.
  • RPM/CPM is the dollar amount earned per thousand impressions that a unit generates. There are many factors that may influence the RPM/CPM, for example: the subject of the content, the time of year, the financial situation and more.
  • Fill Rate is the percentage of paid impressions out of the total impressions. Unfortunately, sometimes there remains unsold inventory, which does not generate income.

The optimizer tracks these metrics and aspires to reach higher CTRs, Fill Rates and CPMs. For example, he will prefer to deliver ads from an ad network that can deliver higher fill rates. Or, he may notice that a specific ad network is generating a better CTR when allocated to business-related content, while the same ad network performs poorly on content geared towards women.

In order to manage this high-scale ad serving, the optimizer uses an Ad Server. This is a smart ad delivery system that lets you define sets of rules for how to deliver the ads, which ad networks go where, priorities, limitations, chaining, and other rules that define the flow of ads in the network. There many vendors of ad servers, like AdvertPro, Adspeed and more. The business they do is usually a basic monthly fee and CPM.

Since all ads traffic goes through the ad server, and since it is the main tool of the optimizer, the ad server should be the natural place for him to read statistics and learn the metrics of the system. Every ad server has its statistics section where you can look at the data from various angles and “cut” it according to all kind of measures. Sounds obvious, right?

Well it appears that there is a little problem: usually there is no data in the reports. Weird but true. The reason is that the ad server needs to speak with third party companies’ code in order to track the activity there, and if it does not “speak” their language, it won’t be able to count and calculate the CTR/CPM/Fill rate and all other metrics that are being generated. So the optimizer has this beautiful Mercedes (Ad server statistics) but no fuel (statistics).

This is a big headache for the optimizer since now he has to log in separately to each ad network and analyze the data there. Although optimizers are very special, trying to manage more than three ad networks is almost impossible. This poor optimizer will need to either develop a memory of a mega computer or be a very skillful Excel user with the copy-paste technique of an experienced programmer, plus the patience of an elderly elephant. Is is true that optimizers are special, but the overwhelming majority of them don’t fit this bill.

Sophisticated Solutions
Companies like Rubicon Project, Pubmatic and ZEDO offer services that do exactly this and more. Not only that you get a detailed dashboard, they also claim to increase revenue by automatically tweaking ad placements, sizes and colors. They run massive A/B tests on your sites, and an algorithm analyzes the results to determine the best ad allocation. This sounds good, but it is not free. They will take their slice of the pie, and it is quite a big slice. Of course as long as they are producing more than their share, it is worth it. They will also need full access to all of the accounts that the media network in question has. Giving the keys to such a sensitive business element does not always leave you with a secure feeling in your gut, but it is doable and things work that way. And again, the important thing is that the optimizer gets access to a well-equipped dashboard that can reveal the big picture, all centralized in one place.

There is another solution, which is also an ad server. Its name is OpenX. The reason why this ad server is different is because it is open source software, so you can download and install it on your server and use it for free. (A hosted option was also added recently). Its story started somewhere back in 1998, and it was re-branded twice before it got its current name. OpenX is a leading ad server that’s open source, and is well-funded ($20.5M) with backing from various venture capital firms. It is stable and reliable, and has all the important features that are essential in a reputable ad server. Instead of paying the traditional ad servers, Optimizers can simply use OpenX.

It is not sophisticated like Rubicon Project, Pubmatic or Zedo. It does not do automatic changes between ads and placements, and it is more like the mainstream ad servers. However, as open source software, it has a huge advantage over the other ad servers – especially if its users are techie and can program, or at least convince someone to program for them.

The idea is to feed OpenX with data from the ad networks, so that when you open the statistics section you see a dashboard full of the valuable data. To achieve this, the structure of the OpenX DB, as well as its code, can be learned.

The feeding system should log in automatically and periodically to each ad network, “read” the relevant data, and feed OpenX with it. Sounds simple, and it is. One of the challenges is to find the relevant data within each report. Each report is different, so it’s necessary to keep some metadata data source that describes how to “read” each report. Again – simple. If the optimizer insists, he can also enhance the basic OpenX reports with more columns and actually create a personalized dashboard: His own dream dashboard.

So optimizers, no more empty reports, no more guessing. Choose OpenX. You will get a free ad serving system that – with a bit of a technological effort – also provides a complete dashboard that will help you increase your revenues and reach your goals.

Related Posts: On this day...

Reader Feedback

One Response to “OpenX: More Than Just an Ad Server”

  1. Elisabeth Giancarlo says:

    Hi there, beautiful site. I couldn’t help noticing the design. Would it be ok if I link to you from my site

Leave a Reply

You must be logged in to post a comment.