Google’s ad server DFP (now called Ad Manager) has long been the most popular ad serving tool in the market, and for a long time, DFP has enabled publishers to add external demand partners to their ad stack. Google did, however, have a trick up its sleeve going in, allowing it to buy impressions before any external demand partners, using Dynamic Allocation in AdX.
This prompted independent ad tech vendors to develop a technology that leveled the playing field for them. This technology is now popularly known as header bidding. In the beginning, header bidding was variously called tagless, pre-bidding, full bidding, parallel auction, and many other names, as reported in this Digiday feature. In June 2018, AdExchanger published this article on header bidding, and the term was cemented. And the rest, as they say, is history.
Header bidding really took off with the publisher community, as you can see with this Google Trends graph that shows the search interest for the term over time.
Despite its widespread adoption and popularity, the technical aspects of header bidding still remain a mystery to many publishers. In order for publishers to understand header bidding, how it works, and its advantages and disadvantages, we’ve created this comprehensive guide.
What is Header Bidding?
Header bidding is a type of programmatic auction in which bid requests are simultaneously sent to multiple demand partners in real-time—which means that every single ad impression has the chance to be purchased at its maximum value, based on available demand. These bids are then passed through various filters (floor price and targeting criteria) configured by publishers in their ad server. The highest bid is selected from the filtered bids, and then finally, the ad creative is displayed on the user’s screen. This entire process takes a fraction of a second.
Why did Ad Tech Need Header Bidding?
Before header bidding, publishers were using the waterfall method of ad serving (also called daisy-chaining) to sell their inventory, under which, the entire inventory is presented to one bidder at a time. The first bidder buys some of the inventory and the remnant inventory is passed on to the next bidder. Similarly, the next bidder buys another part of the inventory, and then again, the remnant inventory is passed to the next bidder in line, and so on.
The publisher waterfall was a really inefficient way of selling inventory. For starters, the inventory took a lot of time to sell, moving from bidder to bidder. Next, publishers were not getting the true value of their impressions. For instance, there were chances that the next bidders in line may have been willing to pay more, but the premium inventory already sold to previous bidders at a lower cost. And there was no way to compare performance and optimize it.
Before header bidding became popular, Google had a monopoly on the ad tech supply chain. Google was (and to some extent, still is) closed to sharing its trade tactics.
Enter header bidding:
- Since the header auction is real-time, it gives publishers the chance to maximize the revenue generated from every single impression.
- Since header bidding sends bid requests simultaneously to all the demand partners, it gives them equal opportunity to make their bids.
- Basis the bids submitted, publishers can compare the performance of each transaction using log-level data and work on tweaks for greater yield.
- It is designed to take precedence over Google auctions. Hence, impressions go to header bidding partners and then to Google’s ad server.
Header Bidding Wrapper and Adapter
Since for each impression, the browser generates an individual bid request to all demand partners, publishers need to insert every demand partner’s code snippet on their website. Inserting or deleting these snippets of code each time a demand partner is added or dropped can get tedious really fast. Also, making constant changes to the header code of a website without dedicated developer support can lead to site outages or the layout being disrupted.
To eliminate these issues (at least to an extent), header bidding wrappers were developed.
A header bidding wrapper works like a container that holds the code snippets from all the demand partners in one place. Some wrappers even offer a GUI, so publishers don’t have to mess around with code at all, instead toggle demand partners on and off from a control panel. Along with that, the wrapper can also be used to set a floor price and the timeouts.
Some well-known header bidding wrappers are Prebid.js, Pubfood.js and BiddR°360.
What is Prebid?
Publishers can allow third-party demand partners to bid on their inventory using the Prebid adapter. So for instance, if AppNexus is one of the demand partners you want to work with, you need to integrate your Prebid with AppNexus’s adapter.
Being an open-source technology, many leading ad tech vendors contribute to the development of Prebid, with AppNexus and Rubicon Project being the two biggest contributors.
Prebid is designed to be flexible and can be used on desktop, mobile, app, and even AMP inventory. It also supports client-side, server-side, and hybrid header bidding (discussed later). Being one of the most popular wrappers, most demand partners support Prebis integrations.
How Does Header Bidding Work?
Header bidding starts as soon as a webpage begins to load on the user’s browser, in parallel, the wrapper initiates the auction by sending bid requests to the demand partners.
Header auction works on the first-price auction model, which means that the highest bidder gets to serve their ad creative, and they pay exactly what they bid during the auction.
Before header auction bids are submitted to the wrapper, the demand partners run their own auctions to decide which advertiser or DSP will participate in the main header auction. Header bidding can be broadly classified into two types: client-side and server-side.
In the case of client-side header bidding, the auction runs on the user’s browser. On the bright side, browser auction offers improved targeting due to better cookie match rates with publishers have access to more bid level data. However, client-side header bidding is notorious for causing high page latency, due to higher browser resource and network consumption.
In the case of server-side header bidding, the auction is conducted on a dedicated auction server away from the user’s browser. This technique saves network bandwidth and browser resources, decreasing page latency as a result. However, when it comes to targeting, the server-side method fails to deliver the desired results due to decreased cookie match rates.
Amazon offers two server-side header bidding solutions—Transparent Ad Marketplace (TAM) and Unified Ad Marketplace (UAM). Where UAM is for small- and medium-sized publishers and TAM is for enterprise publishers.
In order to get the best of both auction types, publishers can also choose hybrid header bidding. This is where publishers run both client-side and server-side header bidding together, usually within a multivariate testing environment.
In case no demand partners’ bid meets the floor price or the timeouts have exhausted, the inventory is offered to fallback networks like AdSense and AdX, based on Ad Manager settings.
Let’s Talk About Video Header Bidding
Just like header bidding works for standard display ads, video header bidding enables publishers to sell their video ad inventory to the highest bidder. Initially, video header bidding required separate web and server integrations, now, many existing wrappers (like Prebid.js) can be configured to sell video inventory, via both client-side and server-side auctions.
Back in 2016, publishers feared that not many buyers were interested in video inventory, hence they may not get the kind of competition seen with display ads. But, in recent years, video inventory has become valuable. According to Statista, video ad spends amounted to $35 million in 2019, and is expected to see a 5% annual growth, reaching $43 million by 2023.
Similarly, as the market for video is growing, many advertisers and demand partners are uniting to encourage publishers to experiment with video inventory. OpenX, AppNexus, and SpotX are some of the demand partners that deal with video ad space.
Effects on Publisher Revenue
In-house example—CCNA7, a niche education website, implemented header bidding with ad refresh to experience 500% revenue uplift in the span of six months.
According to an eMarketer report, the current global spending on header bidding is around $59 billion. And by 2021, this number is supposed to grow by 87% and reach $81 billion.
But, this is one side of data. According to the same research, the growth rate for the adoption of header bidding seems to be decreasing. Right now, header bidding adoption rate is 20% and by 2021, it will become 15%. This is happening because of two things:
- Header bidding is reaching its growth threshold as most target publishers have already adopted the technology.
- Publishers are invested in other competing technologies (like Google’s Open Bidding) for monetizing their inventory.
Whatever the reason, on an individual level, publishers can still make the most of header bidding, as the demand side continues to increase ad spend.
Advantages of Header Bidding
Here are the benefits of header bidding for publishers:
- Global reach: Header auction lets publishers add multiple demand partners. These demand partners are further connected to DSPs and advertisers. This supply chain helps increase bid competition for the inventory and generates more revenue.
- Transparent exchange: Even with w complex supply chain, header bidding is known to keep things really simple. Publishers can easily check which part of the inventory is sold to which demand partner with bid-level data. Also, many wrappers provide advanced reporting and analytics to help publishers monitor their earnings and performance.
- Better control: Header bidding allows publishers to configure the auction as per their needs, this includes customizing the floor price, setting timeouts, and adding or removing demand partners to balance ad revenue with page latency.
- Minimum discrepancy: Since, auction is organized on the publisher’s platform via a wrapper, it helps publishers to track the transactions. Also, running simultaneous auctions helps to reduce the discrepancy—fewer moving parts mean less discrepancy.
- Benefits advertisers too: The success of this tech is made possible because it benefits both the sell- and buy-side. Just like publishers, advertisers are readily purchasing inventory sold via header bidding. The transparent and low discrepancy environment encourages advertisers to invest equally in header bidding.
Disadvantages of Header Bidding
Everything about header bidding sounds great so far. But the technology is not without its challenges. To start with, it is complicated to set up and implement. Next, the client-side method contributes to increased page latency.
Here are some other drawbacks publishers should know about:
- Complicated setup: Implementation process is a bit tricky. From choosing the right wrapper to integrating codes of multiple demand partners, publishers need technical expertise to comprehend and implement header bidding.
- Increased cost: Header auction often requires a dedicated person/s to monitor and improve performance. This includes managing the ad units, adding/removing demand partners, and more. This adds to the cost of running the auction.
- Increased latency: In case of client-side header bidding, the auctions run on the user’s browser, contributing to high page latency. However, with server-side auction like Google’s Open Bidding, at least that particular problem is taken care of.
- Lack of server-side transparency: In case of server-side auctions, things are managed by the vendor running the auction server, making it hard for publishers to access reporting about bid-level and demand partner performance. Lack of transparency makes publishers accept the bids they are getting, minimizing the scope of improvement.
Google’s Entry Into Header Bidding
Google pitched Open Bidding by focussing on its server-side benefits. Basically, it offered a server-to-server auction to publishers. With Open Bidding, Google allowed publishers to bring their own demand and also use Google’s demand to run S2S auctions on Google’s platform.
Both client-side header auction and Open Bidding, have their own pros and cons, so we can’t just say one is better than the other. When compared though, client-side proves to be more transparent and offers better cookie match rates. Whereas Open Bidding saves publishers from complex setup as Google takes care of everything from setup to ongoing auction management.
Publishers can run header bidding and Open Bidding on Google Ad Exchange. This method lets publishers access Google’s demand and add their in-house demand partners to further improve bid pressure.
Here’s how it works:
Once an impression is available, ad requests are sent to Google AdX and header bidding. Within the AdX system, both AdX and Open Bidding buyers place their bids to win the impression. Meanwhile, header bidding starts collecting bid responses from its demand partners. Upon completion, AdX and Open Bidding pass on a buyer(X). Similarly, header bidding passes on another (Y). Finally, AdX compares these bids and choose the winner (highest bid).
How to Get Started With Header Bidding
If the publisher is capable of integrating the website with header bidding wrapper, then Prebid.js is the most suitable option. However, if a publisher lacks the technical expertise to do so, then you may want to opt for a managed service provided by a third-party ad tech vendor.
Once the wrapper is added to the website, publishers to get in touch with demand partners to get access to their respective adapters. The adapters enable publishers to connect with the demand partner’s marketplace and find buyers for their ad inventory.
The setup can be divided into three easy steps:
- Find the right demand partner: Great demand means great revenue. If you have good contacts with ad exchanges and SSPs, then adding demand partner won’t be a problem for you. Don’t just go for the popular partners, examine and compare their services, and choose the ones that suit your audience profile basis the niche you operate in.
- Install a header bidding wrapper: Next, you need to place a header bidding wrapper on the website. This works like a container, where you will be placing codes from all your demand partners and setting auction settings and rules. After this, every time an impression is available, your demand partners will receive requests to place bids.
- Set timeouts and floor price: Adding too many demand partners without configuring the campaign can result in slow page loading. Hence, publishers must setup session timeouts, i.e., time limits for auction termination. Similarly, publishers need to set floor prices for different ad units and add any other limitations, as required.
Once done with steps, you still need to monitor and optimize the setup on an ongoing basis to make sure that you’re maximizing impression-level revenue without causing usability problems. Here are some browser extensions publishers and their ad operation teams can use to optimize header bidding auction:
Headerbid Expert: A Chrome extension powered by Prebid, this tells the number of demand partners bidding on a certain webpage. Further, it tells the time taken by each partner to place their bids. Basis the quick analysis provided the extension, publishers get to know whether their timeout period is considerate or they are underselling their inventories and methods to fix these issues.
BidFilter: This extension shows detailed results like names of the demand partners added in the wrapper along with the bids received on individual ad sizes. The tabs on extension shows list of all bids, winning bids, and empty bids (partners who didn’t bid intentionally or due to timeout).
MyAdPrice: This extensions shows lists of demand partners, winning bids, time taken by teaching partner to respond to the bid request, and total auction time down to each ad size. It also tells the money made by publishers by each visitor under “Website Revenue” and “Average bid prices”.