Auction vs Waterfall

Auction vs Waterfall

Choosing the right ad calling strategy plays a key role in the success of publishers. This choice not only affects revenue growth but also user experience, which in turn can reduce the load on site speed caused by advertising. The two main strategies for selling ad inventory today are auction and waterfall. Auctions allow advertisers to bid while the user's page loads, which usually results in higher revenues for publishers. The waterfall represents a cascading distribution of ad impressions, where inventory is first offered to one advertiser, then to the next, and so on. This can be more predictable but less profitable.

What is Header Bidding?

Header bidding (HB) is a technology that allows publishers to offer their ad inventory to multiple buyers simultaneously. HB creates competition among advertisers and typically increases revenues for publishers. Unlike the waterfall, where bids are made sequentially as codes are called, HB allows all auction participants to bid simultaneously.

Server-side header bidding

It is worth noting that Yandex HB, widely used in Russia, is client-side header bidding, meaning the auctions happen on the client side, and the user can see which bids, for example, were placed on them. There is also server-side HB.

Server-side header bidding is a technology that optimizes the auction process for ad spaces by shifting the load from the browser to the server. Server-side header bidding makes requests to various ad networks on the publisher's server. This reduces page load times and improves performance.

Examples: Prebid Server, Amazon TAM, and Google Open Bidding.

Beyond Russia, the main example of client-side HB is Prebid.js. Unlike the Russian technology controlled by Yandex, Prebid.js is community-owned, and its code is open to everyone. Why Yandex does not do this is a topic for discussion.

Russian header bidding

Advantages of HB

  • High revenue. HB usually increases the average cost per thousand impressions (CPM) and fill rate by involving multiple platforms in the auction.
  • Ease of integration. Integrating partners does not require working with code or embedding one code into another. All changes are made in a fairly straightforward configuration list.
  • Performance. Simultaneous polling of all participants and limiting the auction duration (userTimeout) provide faster ad delivery to the user, improving visibility.

Disadvantages of HB

  • Possible delays. Additional scripts and numerous simultaneous connections can slow down page loading and negatively affect user experience. In this case, it is recommended to disable HB for banners located on the first screen of the page, as speed is crucial for them.
  • Limited number of partners. To work with HB and purchase inventory, a special adapter must be developed. Not everyone can do this, and not everyone has one. However, such code can always be connected to residual traffic.

What is Waterfall?

What is Waterfall?

Waterfall is a more traditional model where ad inventory is sold sequentially by calling the codes of ad platforms one by one. If the first network does not buy the inventory, the request moves to the next network, and so on. This process continues until the inventory is sold or the list of networks is exhausted.

Advantages of Waterfall

  • Load speed for the first code in the "chain." If you are very confident that the first code is highly likely to buy your impression, it might be a good strategy to insert such a platform's code directly without using AdFox and HB, which will significantly reduce load times and overall user load.
  • Prioritization. The waterfall method allows publishers to set clear priorities and task order, which can sometimes be useful.
  • Additional partners who do not develop their own adapter and work only through direct integration. These are usually foreign monetizers who do not want or cannot consider the specifics of a particular market.
Disadvantages of Waterfall

Disadvantages of Waterfall

  • Low revenue. Unlike HB, the waterfall does not provide high competition among advertisers, resulting in lower revenues for publishers.
  • Low ad rendering speed. In the waterfall, each network is processed sequentially, which significantly slows down ad rendering speed on the page, reducing visibility and fill rate.
  • Complexity in setting up various codes and callbacks. In the waterfall, you have to deal with each code separately and set up a callback for each partner. This increases setup and maintenance time.

Ad rendering is the process of displaying ads on a web page or in an app. When a user opens a page, the browser or app sends a request to the ad network server, which selects a suitable ad and sends it back for display. An important aspect and metric is the speed of ad rendering. The higher the speed, the faster the ad will be shown to the user, and the more likely it will be counted and paid for.

Comparison of HB and Waterfall

  • Revenue. HB typically provides higher revenue due to competition among advertisers, while the waterfall leads to lower revenues due to sequential inventory sales. However, in some cases, it makes sense to consider using a short waterfall.
  • Speed. HB can reduce page load speed due to a large number of simultaneous requests and scripts. While the waterfall works slower, sequentially calling each partner in the chain, it minimally slows down and consumes resources.
  • Setup. Once you set up HB, you can simply work. Unlike the waterfall, where changing priorities requires rewriting and testing the entire chain code.
Comparison of HB and Waterfall

Conclusion

Frankly, it is hard to imagine someone removing header bidding from their site and installing a waterfall right now. This may be justified in certain cases, such as when a particular network buys a lot, very fast loading is required, and in other situations, but usually, the auction brings more stable and higher revenue.

Other related materials