The Mechanics Of Real-Time Bidding

Many newcomers to online ad buying have questions regarding the traffic they can buy through a demand-side platform (DSP). How much does it cost? How much of it is available? What is the quality like?

The answers are actually interrelated, and require an understanding of the mechanics behind how real-time bidding (RTB) works.

Industry Terminology Explained

First, a clarification of some industry terminology. The first page that a visitor views — more often than not the home page — is called the “first impression.” The second page is the “second impression,” and so on.

Since most advertisers covet lower session depth (session depth refers to the number of pages viewed by a visitor in a particular viewing session, with lower depth corresponding to earlier views), publishers will often sell these impressions in bulk through direct deals with advertisers. Impressions that are not sold through direct deals (usually those with a higher session depth) are sold through RTB to the highest bidder.

Let’s take the fictional example below:

Three advertisers bid to show their ads to a specific website visitor on a particular site. Now, let’s say that this visitor views nine pages on the site before leaving, and all nine impressions are sold using RTB.

The 1st impression will be sold to the advertiser with the highest bid. Since RTB operates on a “second price” auction model, the amount paid by the winning bidder will always be $0.01 more than the second highest bidder. Now, our website visitors loads a 2nd page on the site.

This 2nd impressions is once again sold to the highest bidder, using a second price auction. This process continues each time our website visitors opens a new page.

Frequency Capping & RTB

But, let’s say that our advertisers have capped the number of times they want to show an ad to a particular visitor. Once that “frequency cap” is reached, they will no longer bid on available impressions. In our example, advertiser A has a cap of three impressions per 24-hour period. So, when our website visitor opens the 4th page on the website, advertiser A will not bid on the impression. This allows advertiser B to win the impression, at a price $0.01 greater than advertiser C’s bid price.

What does this mean for advertisers? For one, the advertiser that bids most will always have their ad shown first. While this may seem intuitive, it has important implications for an advertiser’s bidding strategy. Higher bids should result in lower session depth, and lower bids in higher session depth. But, higher bids should also yield a greater number of impressions.

Some website visitors only open a few pages on a site before leaving, and will never get to the session depth at which a low-bidding advertiser can expect to win an impression (especially if other advertisers have higher, or no, frequency caps). An advertiser with a lower bid price, therefore, can expect to win not only higher session depth impressions, but also fewer of them.

RTB is very fluid and dynamic, and it is difficult (if not impossible) to model the behavior of all other advertisers bidding on a particular impression. But, as shown above, the basic rule of auction bidding appears to hold true: if you want the best quality inventory, and the greatest volume of impressions, bid high!