Most explanations of header bidding are written by the companies selling it. SSPs, managed service providers, ad networks — they all have a version of the "what is header bidding" article, and they all conveniently skip the parts that don't serve their pitch.

This isn't one of those articles.

I've spent over 12 years on the publisher side of programmatic advertising. I've implemented header bidding setups across hundreds of publishers — from sites processing under 100 million monthly ad requests to operations handling over a billion. I've seen what works, what breaks, and most importantly, what your current ad partner probably isn't telling you about how your inventory actually gets sold.

By the time you finish reading this, you'll understand what header bidding is, how it works mechanically, why it matters to your bottom line, and — this is the part most guides leave out — what to watch for so you're not leaving money on the table while thinking everything is fine.

How Publishers Used to Sell Ad Inventory (And Why It Was a Problem)

Before header bidding, publishers sold their programmatic ad inventory through a system called the waterfall — also known as daisy chaining. If you're still running one, you'll recognize this immediately.

Here's how it worked. Your ad server would call demand sources one at a time, in a fixed order. The first source in the chain got the first shot at your impression. If they didn't want it (or didn't meet your floor price), the request moved to the next source, then the next, and so on down the line.

The problem? The order of that chain was rarely based on who would actually pay the most for a given impression. It was based on historical averages, relationship dynamics, or whoever negotiated the best position with your ad ops team. A demand partner sitting in position four might have been willing to pay three times more than the partner in position one — but they never got the chance to bid because someone else already filled it at a lower price.

Waterfall vs Header Bidding: Same Impression, Different Revenue

Three SSPs want to bid on your ad impression. Watch what happens under each model.

Waterfall Model
Sequential — one buyer at a time
#1
SSP Alpha
Asked first — fills at floor price
€2.00 Wins
#2
SSP Beta
Never gets asked — impression already filled
€4.50 Skipped
#3
SSP Gamma
Never gets asked — impression already filled
€3.20 Skipped
€2.00
SSP Beta would have paid €4.50 — but never got the chance. You left €2.50 on the table on this single impression.
Header Bidding
Simultaneous — all buyers compete at once
SSP Alpha
Bids in real time
€2.00
SSP Beta
Bids in real time — highest offer
€4.50 Wins
SSP Gamma
Bids in real time
€3.20
€4.50
Every SSP competed. The highest bidder won. That's +125% more revenue on the same impression, same inventory, same user.
💡
This example shows a single impression. Now multiply that difference across millions of daily impressions. That's why well-configured header bidding typically delivers a 20–40% revenue uplift over pure waterfall setups.
programmatic.expert · Gediminas Blažys

The waterfall model was predictable and easy to manage. It was also quietly costing publishers significant revenue. I go deeper into this in my comparison of header bidding vs waterfall, but here's the short version: the waterfall was designed for simplicity, not for maximizing your yield.

Header bidding was the industry's answer to that problem.

Header Bidding Explained: What It Actually Is

Header bidding is an advanced programmatic technique that lets multiple demand sources bid on your ad inventory simultaneously — before your ad server makes its decision.

Instead of asking buyers one at a time whether they want to buy an impression (the waterfall), header bidding asks everyone at once. All participating demand partners — SSPs, ad exchanges, ad networks — receive the bid request at the same time, submit their bids within a defined time window, and the highest bid gets passed to your ad server to compete with any direct campaigns or other demand you have set up there.

Think of it this way. The waterfall was like selling your house by asking one potential buyer at a time if they'd meet your asking price. Header bidding is an open auction where every interested buyer places their best offer simultaneously. Which approach do you think gets you a better price?

The "header" in header bidding refers to the website's HTML header, where the technology was originally implemented. A small piece of JavaScript code — known as the header bidding wrapper — runs in the background every time a user loads your page. It fires off bid requests to your configured demand partners, collects their responses, identifies the winning bid, and passes that information to your ad server.

This all happens in milliseconds, before your page fully renders.

How Header Bidding Works: The 6-Step Flow

Let me walk you through what actually happens between a user clicking on your site and an ad appearing on their screen. Understanding this flow is essential if you want to have meaningful conversations with your ad tech partners about your setup.

From Page Load to Ad Served: 6 Steps

What happens between a user clicking your site and an ad appearing — in milliseconds.

1
Page Loads
User arrives on your site. The header bidding wrapper (e.g. Prebid.js) embedded in your page's code fires up before content fully renders.
t = 0ms
2
Bid Requests Sent
The wrapper sends simultaneous bid requests to all configured demand partners (SSPs) with ad slot size, position, and user consent data.
t ≈ 50ms
3
SSPs Respond
Each SSP evaluates the impression and responds with a bid price — or doesn't. All within the timeout window (typically 1,000–3,000ms).
t ≈ 200–1500ms
6
Ad Served
The ad server makes the final call and serves the highest-paying ad to the user. Total elapsed time from page load: milliseconds.
t ≈ 1500–3000ms
5
Competes in Ad Server
The winning HB bid enters your ad server (Google Ad Manager) where it competes against direct campaigns, AdSense, and exchange demand.
⚠️ "Last look" happens here
4
Top Bid Selected
The wrapper collects all bids, identifies the highest one, and packages it as a key-value targeting pair for the ad server.
t ≈ 1500ms
🔍

Insider Note: Google's "Last Look" Advantage

At Step 5, Google Ad Exchange gets to see the winning header bidding price before deciding whether to outbid it. This dynamic directly affects your revenue and is rarely mentioned in vendor-written guides. It's one of the reasons why understanding your full auction flow — not just your header bidding setup — matters.

Bid collection phase
Winner selection
Ad server decision
programmatic.expert · Gediminas Blažys

Why Header Bidding Matters for Publisher Revenue

The business case for header bidding comes down to three things: competition, transparency, and control.

More Competition Per Impression

In a waterfall setup, a given impression might see bids from two or three demand sources before it gets filled. With header bidding, that same impression might see bids from six, eight, or even ten sources simultaneously. More competition means higher clearing prices. It's basic auction economics.

From my experience implementing header bidding across publishers of varying sizes, a well-configured setup typically delivers a 20–40% increase in programmatic revenue compared to a pure waterfall. The exact number depends on your traffic profile, your geographic audience, your vertical, and critically — how well the setup is actually optimized.

Real Transparency Into Demand

With header bidding, you can see exactly who bid on your inventory and how much they offered. This is enormously valuable data. It tells you which demand partners are actually competing for your audience, which ones are consistently winning, and which ones are barely showing up.

This kind of visibility is something the waterfall model never gave publishers. And frankly, not every managed partner is eager to give it to you now. If you can't see bid-level data from your setup, ask yourself why.

Reduced Dependency on a Single Partner

Header bidding structurally reduces your dependency on any one demand source. When multiple SSPs are competing simultaneously, no single partner has disproportionate control over your revenue. This is a significant strategic advantage — it gives you negotiating leverage and protects you against sudden revenue drops if one partner changes their algorithm or business terms.

At a Glance: Waterfall vs Header Bidding

How the two models compare across the metrics that matter most to publishers.

Waterfall
The legacy approach
Auction Sequential — one buyer at a time
Revenue Limited by priority order
Transparency Low — hard to see true value
Dependency High — one partner dominates
Complexity Low — simple to manage
Uplift Baseline
Header Bidding
The modern standard
Auction Simultaneous — all compete at once
Revenue Maximized by open competition
Transparency High — full bid visibility
Dependency Low — partners compete equally
Complexity Moderate — needs monitoring
Uplift +20–40% vs waterfall
Waterfall
Header Bidding
Auction
Sequential
Simultaneous
Revenue
Limited by priority
Open competition
Transparency
Low
Full visibility
Dependency
One partner dominates
Partners compete equally
Complexity
Simple
Needs monitoring
Uplift
Baseline
+20–40%
💡
Properly optimized header bidding typically delivers +20–40% revenue compared to waterfall setups — but only when someone is actively managing the configuration.
programmatic.expert · Gediminas Blažys

What Your Ad Partner Probably Isn't Telling You About Header Bidding

This is the section you won't find on your SSP's blog. These are the things I've learned from sitting on the publisher side for over a decade — watching how header bidding gets implemented, managed, and sometimes quietly mismanaged.

"We have header bidding" doesn't mean it's optimized

Having header bidding enabled and having header bidding properly optimized are two very different things. I've audited publisher setups where header bidding was technically running but configured so poorly that it was barely outperforming the old waterfall. Wrong timeout settings, outdated adapter versions, missing demand partners, incorrect floor prices — any of these can silently drain your performance.

If your partner set up header bidding for you two years ago and hasn't touched the configuration since, I can almost guarantee it's underperforming.

💡 Key takeaway: Header bidding is not a set-it-and-forget-it technology. The gap between "enabled" and "optimized" is where significant money lives — or dies.

"Managed header bidding" often means managed for their benefit

Many SSPs and ad tech vendors offer "managed header bidding" as a service. That sounds convenient — they handle the complexity, you collect the revenue. But here's what "managed" often means in practice: they choose which demand partners participate in your auction, and unsurprisingly, they tend to prioritize their own demand or their commercial partners.

Ask your managed partner a simple question: which demand partners are in my header bidding setup, and why those specific ones? If they can't or won't give you a clear answer, that's a signal worth paying attention to.

Timeout settings are a hidden revenue lever

Every header bidding setup has a timeout — the maximum time the wrapper will wait for bids before moving on. This setting is usually between 1,000 and 3,000 milliseconds. It seems like a technical detail, but it has direct revenue implications.

Set the timeout too short, and high-value bids from slower SSPs get cut off before they arrive. Set it too long, and your page load suffers, hurting user experience and potentially your Core Web Vitals scores. The optimal timeout depends on your traffic, your demand partners' response times, and your audience's tolerance for latency.

🔍 Insider note: Your ad partner chose your timeout setting. Do you know what it is? Do you know why they chose that number? In my experience, most publishers can't answer these questions — and the answers often reveal optimization opportunities worth exploring.

The gap between "enabled" and "optimized" is where money lives

Header bidding is not a set-it-and-forget-it technology. The programmatic landscape shifts constantly — SSPs change their algorithms, new demand partners emerge, bid patterns fluctuate seasonally, privacy regulations alter targeting capabilities. A setup that was optimal six months ago may be leaving significant revenue on the table today.

The publishers who get the best results from header bidding treat it as an ongoing optimization discipline, not a one-time implementation. If no one is actively managing your setup — reviewing bid data, adjusting configurations, testing new partners — you're almost certainly not getting what you could be. And sometimes, the cause of a revenue decline isn't even inside the ad stack — it can come from changes to your consent management, site layout, or UX that nobody on the ad ops side was aware of.

Is Header Bidding Right for Your Publisher Site?

Not every publisher needs header bidding, and I believe in being honest about that.

Quick Assessment: Do You Need Header Bidding?

A quick decision framework based on your monthly ad request volume — from someone who's helped publishers at every scale.

📊
Small Publisher
< 20M monthly ad requests
Consider Carefully
The complexity and overhead of a full header bidding setup may outweigh the revenue gains at this scale. Your focus should be on getting the fundamentals right first.
Start with a well-configured ad server + 1–2 strong demand partners. Revisit header bidding as your traffic grows.
🚀
Mid-Size Publisher
100M – 500M monthly ad requests
Strongly Recommended
The sweet spot for header bidding ROI. You have enough volume for meaningful competition between demand partners, and the transparency gains help you make better strategic decisions.
If you don't have HB yet, you're almost certainly leaving 20–40% of programmatic revenue on the table.
🏢
Large Publisher
500M+ monthly ad requests
Essential
At this scale, even a 1–2% yield improvement translates to significant absolute revenue. Not having optimized header bidding means substantial, ongoing money loss.
You need HB + continuous optimization. A setup that hasn't been tuned in 6 months is already underperforming.
💡
Traffic volume is the starting point, not the whole picture. The right answer also depends on how much control you want over your monetization, your technical capacity, and whether your current partner's incentives are truly aligned with yours. Header bidding gives publishers more control — but only if someone is actually paying attention to how it's configured.
programmatic.expert · Gediminas Blažys

Regardless of where you fall, the most important thing is to understand what's happening with your ad inventory. Whether you run header bidding yourself, use a managed service, or are still on a waterfall — make sure you understand the setup, the economics, and the incentives of everyone involved.

Frequently Asked Questions About Header Bidding

What is the difference between header bidding and real-time bidding (RTB)?

Real-time bidding is the broader mechanism of buying and selling ad impressions through instant auctions. Header bidding is a specific technique within that ecosystem — it's the method by which publishers allow multiple demand sources to participate in the RTB auction simultaneously, before the ad server makes its final decision. All header bidding involves RTB, but not all RTB involves header bidding.

Does header bidding slow down my website?

It can, if it's poorly configured. Each demand partner added to the wrapper adds potential latency. However, a well-optimized setup with proper timeout management and the right number of partners has minimal impact on page load times. The trade-off between a few hundred milliseconds of latency and significantly higher ad revenue is almost always worth it — as long as someone is actually monitoring and optimizing that balance.

Should I use client-side or server-side header bidding?

Both have tradeoffs. Client-side runs in the user's browser and offers better targeting (because of cookie access) but adds latency. Server-side runs on a remote server and is faster but sacrifices some targeting capability and transparency. Most publishers I work with run a hybrid — top demand partners client-side, the rest server-side.

How many demand partners should I have in my header bidding setup?

There's no universal number, but more isn't always better. Each additional partner adds latency and complexity. The optimal number depends on your traffic volume, geographic audience, and the unique demand each partner brings. For most mid-size publishers, five to eight well-chosen partners is the sweet spot. The key word is "well-chosen" — five strong, unique demand sources beat twelve overlapping ones.

How much does header bidding cost to implement?

If you're using an open-source solution like Prebid.js, the technology itself is free. The cost comes from the expertise needed to implement, configure, and maintain it — either in-house talent or through a partner. Managed header bidding services typically take a revenue share, usually between 5% and 20% of the programmatic revenue they manage. Whether that's a fair deal depends entirely on the value they're adding. If you're paying 15% revenue share and your setup hasn't been optimized in a year, that's a question worth raising.

Not sure if your header bidding setup is working as hard as it should? I offer a free 30-minute diagnostic call where I'll tell you what I see — what's working, what's not, and what I'd change. No pitch, no strings, no vendor agenda. Just an honest look from someone who's been on your side of the table for over a decade.

Book a free diagnostic call