Evolution of Digital Ad Serving — Part I

Pandurang Patil
5 min readAug 17, 2015

In the early days when there were no AdServers to show dynamic ads, publishers used to embed ads statically along with their content. So to display a new ad in a given slot, the publisher has to change their content every time to embed the new ad. Moreover, every user will see the same ad at a given instant of time.

Picture — 1

With the growing trend to utilize digital space for advertising, it becomes more obvious to utilise power of digital world for advertisement, as it gives more control over how ads can be displayed. It is easier to get measurable direct feedback that can be utilized to optimize campaigns, which is very difficult with other mediums of advertisement (Print, TV, and Radio), even today. Therefore one needs to have an intelligent system that could dynamically consider this feedback and decide which ad to serve for a given user at runtime. This is where AdServer comes into the picture. In the early days publishers were using ad servers like DART Enterprise (by DoubleClick now acquired by Google Inc. At present more companies provide publisher AdServers e.g. DFP ( DART for Publishers), AdZerk etc. Now Agnie Media Software also provides a Publisher Ad Server ). AdServer enables publishers to manage advertisers’ campaigns to show ads dynamically on their portals. In this case, publishers have their own channels to get the direct ads.

To integrate with AdServer in place of direct static ads, the publisher now uses Ad-Tag provided by AdServer. This Ad-Tag fetches ads from AdServer dynamically in real-time.

Picture — 2.

With the introduction of AdServer in the market, publishers were dynamically able to show different ads from different campaigns. However, with time they started realizing they didn’t have sufficient direct demand (Advertiser Campaigns) to show ads round the clock and fill all the ad slots. As many publishers started facing this issue, it gave way for the mediator to rise between publishers and advertisers, which we know as Ad Network. To integrate with Ad Network, publisher AdServer provides a mechanism called pass back / Default Ad-Tag. Publishers need to register with Ad Network to get their Ad-Tag for an Ad slot and configure the same Ad-Tag as a pass-back/default tag in the publisher Ad Server. This way in the event a publisher AdServer doesn’t have any ad/campaign to serve through direct channels, it passes this opportunity to Ad Network to serve ads. In this situation even if Ad Network doesn’t have any ad to serve, the publisher can pass on the opportunity to another AdNetwork using a similar mechanism. This way a publisher can create a chain of Ad Networks and monetize unsold/blank ad requests through other networks. However, the price will be much lower than the ads that the publisher gets through direct channels.

You might have heard of Guaranteed Inventory and Remnant Inventory, let's understand what is it. The ads that are displayed as a result of direct Advertiser campaigns on the Publisher’s website are called Guaranteed inventory. The ads that are displayed after the Guaranteed inventory has consumed their ad slots through the third-party Ad Networks are called Remnant Inventory

Picture — 3.

With the presence of multiple Ad Networks in this space, publishers have different revenue models with different Ad Networks. Each Ad Network pays different pricing for a given ad request, so if a Publisher wants to generate more revenue they must pass the very first opportunity to an Ad Network which will pay more. However, the challenge publishers started facing was that one Ad Network consistently was not able to provide higher revenue. Dynamics change with time, as well as the demand (Advertiser campaigns) an Ad Network has at a given time. These parameters keep on changing over the period of time, which results in a situation where at one time “Ad Network 1” pays more and the other time “Ad Network 2”, and some other time any other “Ad Network” will pay more. In order to maximize revenue, a publisher had to keep on monitoring the revenue being generated through different Ad Networks and keep on changing the pass back chaining to provide the first opportunity to an “Ad Network” that pays more for a given duration. This way Publishers have to apply a good amount of human resources to do all this manually.

What if this process of manual work is automated? This new automated system will process the historical and current data of revenue generated through different Ad Networks and dynamically decide which Ad Network will get an opportunity to serve the first ad or subsequent ads. This system is nothing but SSP (Supply Side Platform) like PubMatic, Rubicon Project, and Ad Meld. SSP acts as a Revenue optimizer for Publishers by dynamically predicting which Ad Network will pay more for a given ad request. To integrate with SSP, the publisher registers with the SSP and then they will provide an Ad-Tag for a given Ad Slot. Publishers configure this Ad-Tag as pass back Ad-Tag in their Publisher Ad Server in place of Ad Network ad tag. SSP gets an opportunity in the event Publisher Ad Server doesn’t have any direct ads to serve. SSP will select an Ad Network that will pay more as per their prediction, and hence lift the publisher’s revenue.

Picture — 4

Now the revenue optimization problem of the Publisher becomes the problem statement for the SSP and they started exploring more ways through which the revenue of publishers can be lifted. Meanwhile, RTB (Real Time Bidding) in Digital Advertising was taking traction in Ad Exchanges (Will talk about Ad Exchange in a separate article). SSPs also introduced the RTB feature in their platform which in turn started providing additional revenue lifts to the publishers.

Originally published at blog.agnie.net.

--

--