What is header bidding, in one sentence? Prebid.org, the open-source header bidding project, puts it this way: header bidding “enables publishers to capture bids for ad units from demand sources that might otherwise have been missed.” The key word is “missed”. Under older setups, buyers who might have paid more never got the chance to bid.
This post follows Prebid’s own documentation, “Introduction to Header Bidding”, checked on 4 October 2026. It does not give revenue figures, because Prebid does not publish any and no honest general figure exists. What it can give you is the mechanism, which is what you need to judge a network’s pitch.
What is header bidding, step by step
Prebid describes the sequence in six stages. Nothing in it is exotic; the important part is the order.
- The page starts to load.
- Header bidding code on the page asks a set of demand partners for bids.
- The demand partners reply with their bids.
- The ad server is called, with some or all of those bids attached.
- The ad server considers all available ads, including the header bids.
- The ad server picks the winner for each ad unit.
The shift is in stage four. Bids arrive at the ad server as real prices from competing buyers, rather than the ad server working through a fixed list of networks on its own terms. Every partner gets to compete for every impression at once.
Header bidding vs the waterfall
The setup header bidding replaced is usually called the waterfall. In a waterfall, the publisher’s ad server calls networks one at a time in a set order. The first network gets first refusal; if it passes, the impression drops to the next, and so on down the list. The order is set in advance, typically on expected value, not on what each buyer would actually pay for this impression.
The weakness follows from that. A buyer further down the list might have paid more for a particular impression, but never saw it because an earlier network took it first. That is exactly the “might otherwise have been missed” demand Prebid describes.
The waterfall persisted for a reason: it is simple. One ad server, a ranked list of networks, no extra code running in the page head. For a publisher with one network and modest traffic, that simplicity still has value. Header bidding earns its complexity only when enough buyers want your inventory for their competition to change the price. If only one or two would bid, running them in parallel adds requests without adding much tension to the auction. That is why the question of who your bidders are comes before the question of how they are asked, and why a network’s list of demand partners tells you more than its description of its technology.
| Waterfall | Header bidding | |
|---|---|---|
| How buyers are asked | One after another, in a fixed order | All at once, before the ad server decides |
| What the ad server sees | Whichever network accepts first | Some or all header bids alongside its other ads |
| Competition per impression | Limited by the order | More competitive, per Prebid |
| Publisher visibility | Low | More transparency and control, per Prebid |
| Cost to the page | Sequential calls | “Some latency in loading the page and displaying the ads” |
Source: Prebid.org, “Introduction to Header Bidding”. The waterfall column describes the general model.
The benefits Prebid claims, and what they mean
Prebid lists two benefits: more competitive auctions, and transparency and control for publishers. Both are modest claims, and that is to its credit.
More competitive auctions means more buyers see each impression at the same moment. It does not promise a particular uplift. Whether competition raises your revenue depends on who is bidding, for which readers, in which month. A site with a narrow, low-value audience may find extra bidders add little.
Transparency and control are easier to value. When bids from named partners reach your ad server, you can see who bids, how often and at what level, and you can drop a partner that adds latency without adding much demand. In a waterfall, most of that is hidden inside each network.
The cost: latency on your page
Here the received wisdom needs correcting. Header bidding is often pitched as free money. Prebid’s own documentation is more honest: “This can cause some latency in loading the page and displaying the ads.” The page waits on bidders before the ad server can decide, and every extra partner is another request.
That trade matters for search. A slower page can cost readers before it earns from them, and page experience is something you can measure. Our guide to page speed and SEO covers where speed affects rankings and where it does not. The practical point: the number of bidders is a setting, not a virtue. More is not automatically better.
How small publishers actually get header bidding
Running header bidding yourself means installing a wrapper, signing agreements with demand partners, configuring an ad server, and tuning timeouts. That is real ad operations work. Most small and mid-sized publishers do not do it. They join a managed network that runs header bidding on their behalf and takes a share of revenue for it.
That moves the question from “should I use header bidding?” to “which network’s setup and terms suit my site?” Entry thresholds vary. Our posts on Mediavine requirements and Raptive requirements cover two of the better known managed options, and the ad network comparison sets the wider field side by side.
When you assess a managed network, ask about the mechanism rather than the promise. Which demand partners bid on your inventory? What is the timeout before the ad server stops waiting? Can you see bid-level reporting, or only a total? A network that answers those questions clearly is using header bidding’s transparency for you rather than keeping it for itself.
One limitation to keep in view. Prebid’s documentation explains how header bidding works and what it is for, but it does not tell you how much more you will earn, and nothing general can. The outcome depends on your audience, your bidders and how much latency your pages can absorb. Treat any uplift figure in a sales pitch as that network’s claim, test it on your own pages, and judge it over several months rather than one.
Frequently asked questions
What is header bidding in simple terms?
Code on the page asks several demand partners for bids at once, then passes some or all of those bids to the ad server, which picks the winner for each ad unit.
Is header bidding better than the waterfall?
Prebid says it makes auctions more competitive and gives publishers more transparency and control. It also admits it can add latency, so it is a trade rather than a free gain.
Does header bidding slow down a website?
It can. Prebid’s documentation says it “can cause some latency in loading the page and displaying the ads”. The number of bidders and the timeout setting affect how much.
Do I need to set it up myself?
Usually not. Most smaller publishers get header bidding through a managed ad network rather than running a wrapper and partner contracts themselves.
The takeaway Header bidding lets every buyer compete for each impression at once, which beats a fixed waterfall on competition and visibility. It costs page speed, so judge any setup by its bidders, its timeout and its reporting, not by a promised uplift.
