Since its emergence in the early 2000s, programmatic advertising has experienced rapid growth. Automating the buying and selling of advertising space has enabled publishers to make their inventory more profitable, and advertisers to optimize their advertising investments.
But over the past twenty years, this marketing method has had to evolve in depth to overcome certain technical limitations. From its initial waterfall structure to the rise of header-bidding, the market is now evolving towards a new stage: server-side programmatic.
In the beginning: a waterfall approach
Programmatic advertising began as a waterfall model. In this context, available ad impressions are offered to a series of partners (adnetworks or SSPs) one after the other. The order of priority is defined by the publisher or its adserver according to various criteria, such as the average CPM and performance history of each partner.
If the first partner on the list declined the impression, or if its bid was not high enough (floor price), it was then offered to the next partner…and so on until the impression was sold or expired.
The waterfall model is perfectly suited to the imperative of maximizing the fill rate of publishers’ advertising space, but it suffers from one major limitation: the absence of direct competition between buyers. This means that impressions are sold, but not necessarily at the best price. Indeed, SSPs with the highest bids may not be called if a buyer further up the chain has already made the sale.
Other limitations have gradually weakened this model:
- Rigidity: « sequences » are managed manually, which means they cannot be dynamically adapted to the offer.
- Latency: each print refusal triggers a new server call to move on to the next partner.
The rise of header-bidding
In the early 2010s, the limitations of the cascade system paved the way for a new, more competitive and dynamic model: header-bidding. This approach allows all partners to compete simultaneously, via a tag integrated into the header of web pages consulted by Internet users. This equitable access to impressions means that the best bid – not the first bid – wins the auction, boosting the publisher’s revenues.
However, this method comes at a considerable technical cost. Indeed, client-side header bidding makes intensive use of the resources (bandwidth) of the Internet user’s terminal. Already a constraint on the desktop, this load has become critical in the mobile-first era.
To preserve the performance of their sites, publishers have no choice but to limit the number of programmatic partners called for each impression. But this automatically reduces competition…and therefore revenues.
Server-bidding: the new programmatic era
Server-side bidding was introduced to replicate the benefits of header bidding, while eliminating its negative effects on performance. This new method centralizes advertising requests into a single call to a server. The server then simultaneously queries all the partners selected by the publisher.
And the benefits are numerous:
- Drastically reduced latency and improved user experience;
- Ability to call a larger number of SSPs simultaneously without impacting site performance;
- Increased opportunities to sell at the best price, thanks to increased competition.
- Improved carbon footprint of programmatic flows by reducing energy consumption for both publisher and user.
Server-side thus enables publishers to set up truly fair competition between buyers, while controlling the technical performance of their site. The latter is an increasingly crucial issue in the age of Core Web Vitals and Google’s SEO requirements.
The move towards server-bidding marks a strategic step in the evolution of programmatic advertising. This model enables publishers to combine technical performance, fair competition, and greater advertising profitability. As mobile, privacy, and site speed become top priorities, the programmatic cloud is emerging as a modern, sustainable response to the challenges of digital advertising.
