Real Time Bidding Explained: How RTB Auctions Work

Most programmatic advertisers set a bid, launch a campaign, and assume the system does the rest. However, what actually happens inside RTB auctions is far more precise than that. Real time bidding resolves each impression in under 100 milliseconds. The outcome depends on bid request data, floor prices, clearing mechanics, and competitor activity that most advertisers never see.

Understanding real time bidding at a surface level is one thing. Understanding how RTB auctions work at the mechanics level is what separates teams that optimize effectively from those that simply spend. According to Future Market Insights via Xapads, the global RTB market will reach $26.32 billion in 2026, up from $21.02 billion in 2025 at a CAGR of 25.2%. At that scale, every misaligned bid multiplies across billions of auctions.

This guide covers everything. You will learn what RTB is and how each auction step works. You will also understand the role of DSPs and SSPs, how floor prices and bid shading affect costs, and how to use auction data to improve performance. For the broader context of how RTB fits into programmatic advertising, start with our complete programmatic guide.

What Is Real Time Bidding?

Specifically, real time bidding is an automated auction mechanism. In practice, it lets advertisers buy individual digital ad impressions one at a time, as a user loads a webpage or opens an app. Notably, the auction resolves before the page finishes loading. According to AI Digital, each auction completes in 40 to 120 milliseconds. That is faster than a human blink. During that window, the system evaluates who the user is, what the page is about, and how much each advertiser is willing to pay. Ultimately, the winning bid determines which ad appears.

In particular, RTB is a subset of programmatic advertising, not a synonym. Not all programmatic deals use real time bidding. Private marketplace deals, preferred deals, and programmatic guaranteed contracts are also programmatic. However, RTB is the most widespread form. It accounts for the largest share of global programmatic transactions. Indeed, the advantages of programmatic advertising platforms are most clearly visible in the RTB model, where automation, targeting, and pricing all happen simultaneously.

Where Real Time Bidding Came From

The Problem with Manual Ad Buying

Before RTB, digital advertising operated similarly to traditional media. Specifically, publishers listed their ad inventory at fixed prices. Instead of competing in auctions, advertisers negotiated rates and booked placements weeks in advance. As a result, the process was slow. It required phone calls, emails, and signed contracts. Publishers typically sold their premium placements to large brands and passed leftover inventory to ad networks at deeply discounted rates. Advertisers, meanwhile, paid the same price for every view regardless of who was actually looking.

This model had two core problems. First, publishers consistently left money on the table. Premium audiences were undervalued in bulk deals. Second, advertisers had no mechanism to bid more for a high-value user and less for a low-value one. Consequently, both sides operated on historical averages rather than real market signals. RTB was built specifically to fix this.

How RTB Changed the Model

Real time bidding replaced fixed pricing and bulk buying with per-impression auctions. Now, each time a user loads a page, the publisher’s system creates an auction for that specific impression. Multiple advertisers compete simultaneously. The highest qualifying bid wins. The entire sequence takes milliseconds.

This change benefited both sides. Publishers now earn the true market value for each impression. Advertisers pay only for impressions that match their targeting criteria. Additionally, RTB introduced transparency that manual buying never had. Advertisers could see exactly which placements performed and adjust bids accordingly. Publishers, in turn, could see precisely how much demand existed for their inventory at any given moment.

How RTB Auctions Work: The Step-by-Step Process

Every RTB auction follows a precise sequence of events that begins the moment a user loads a page and ends before that page finishes rendering. Here is how each step works.

Step 1: The Bid Request Is Generated

When a user opens a webpage, the publisher’s ad server detects an available ad slot and triggers a bid request. This request travels from the SSP to the ad exchange, which broadcasts it to all connected DSPs simultaneously. Any DSP that fails to respond within the timeout window, typically 80 to 120 milliseconds, is disqualified from that impression entirely.

Bid requests follow the OpenRTB protocol, the IAB-published specification that standardizes communication between exchanges and DSPs. Key fields include the impression object, the site or app object, the device object, and the user object (anonymized audience signals). Additionally, the request includes the publisher’s floor price: the minimum CPM below which no bid is accepted.

Step 2: DSPs Evaluate and Bid

Each DSP receiving the bid request evaluates the impression against its active campaigns in real time. The DSP asks several questions simultaneously: does this impression match any campaign’s targeting parameters? What is the estimated value of showing an ad to this user in this context? What bid price balances winning the impression against paying more than necessary?

If the impression matches, the DSP calculates a bid and returns it to the exchange within the timeout. If no campaign matches or the calculated bid falls below the floor price, the DSP returns no bid. Consequently, DSPs that miss the timeout window are excluded from the auction entirely, regardless of what they would have bid. Latency in bid processing therefore directly affects a DSP’s ability to compete for inventory.

Step 3: The Exchange Clears the Auction

Once bids arrive, the exchange collects all responses within the timeout window and applies its clearing logic. The highest valid bid above the floor price wins. The exchange notifies the winning DSP and delivers the ad creative to the publisher’s SSP, which renders it for the user. All of this completes before the page finishes loading.

Floor Prices: The Publisher’s Minimum Threshold

Floor prices are the lowest CPM a publisher will accept for an impression. Bids below the floor are rejected even if they are the highest bid received. According to Fraudlogix, well-calibrated floor prices maximize publisher revenue by balancing fill rate against CPM yield. A floor too high reduces fill rate. A floor too low leaves money on the table.

Modern SSPs do not use static floor prices. Instead, they apply dynamic floor pricing that adjusts based on real-time signals: time of day, user segment value, content category, and current demand levels. Consequently, an identical ad slot on the same website can carry different floor prices at different moments. For advertisers, this means a static bid strategy produces inconsistent win rates even on familiar inventory.

Floor prices also interact with bid shading algorithms. A DSP shading its bids too aggressively may fall below a dynamically raised floor and lose an impression it could have won at a reasonable price. Well-calibrated shading systems monitor floor price signals to avoid this outcome.

First-Price vs. Second-Price Auctions

Second-Price Auctions: The Legacy Standard

In a second-price auction, the highest bidder wins but pays only the second-highest bid plus one cent. This model historically encouraged honest bidding: advertisers could bid their true value without fear of overpaying. If Bidder A submitted $5.00 CPM and Bidder B submitted $3.00 CPM, Bidder A would win and pay $3.01. According to Clearcode, this gap between bid and clearing price was known as the consumer surplus. However, publishers and SSPs routinely manipulated floor prices and applied undisclosed fees, eroding this advantage.

First-Price Auctions: The Current Standard

The industry has fully migrated to first-price auctions. The highest bidder wins and pays exactly what they bid. According to Setupad, Google completed its transition in 2019 and most major exchanges followed. In a first-price environment, the incentive to bid your true maximum value disappears. Bidding $5.00 and winning against a $3.00 competitor means paying the full $5.00 rather than $3.01. Consequently, bid calibration becomes far more important than it was in the second-price era.

Bid Shading: Controlling Costs in a First-Price World

Bid shading emerged as the direct response to first-price auction adoption. It is an AI-driven technique that reduces a DSP’s submitted bid toward the estimated clearing price rather than the advertiser’s theoretical maximum. Without bid shading, advertisers routinely win impressions at their full maximum bid when a lower price would have secured the same win.

Specifically, bid shading algorithms analyze historical auction data for similar impressions, competitor bid patterns, and floor price signals. They calculate the minimum bid likely to win and submit that price. According to Xapads, bid shading is now a non-negotiable capability in open exchange RTB. Platforms without it systematically disadvantage their advertisers.

However, bid shading also introduces risk. If a shading algorithm reduces bids too aggressively, it sacrifices win rate on impressions the advertiser actually needed. Well-designed shading systems therefore balance cost efficiency against campaign delivery goals. This balance is calibrated dynamically rather than fixed as a flat discount. In practice, effective bid shading reduces average CPMs without sacrificing win rates on target impressions.

Win Rate: The Diagnostic Metric for Bidding Accuracy

Win rate is the percentage of auctions a DSP wins relative to the total it entered. According to Xapads, a healthy win rate for performance campaigns sits between 15% and 40%. A sustained win rate below 10% indicates underbidding or over-targeting competitive inventory. A win rate above 60% typically indicates overbidding, meaning the advertiser is consistently paying more than necessary to win.

Monitoring win rate alongside CPM and conversion data allows continuous bid calibration. For example, if win rate drops while CPMs hold steady, the likely cause is increased competition from other buyers. If win rate stays high but conversion rates fall, the issue is targeting rather than bid pricing. In this way, win rate functions as a diagnostic tool for auction-level campaign health rather than a vanity metric.

Private Marketplace Auctions vs. Open Exchange RTB

Not all RTB auctions operate under the same rules. Open exchange RTB is accessible to any qualified DSP with a bid request. In contrast, private marketplace (PMP) auctions are invite-only. Publishers curate a specific buyer set who receive access to premium inventory before it reaches the open exchange. Deal IDs tag these private transactions, allowing the exchange to apply different pricing rules and priority logic to deal-tagged bids versus open market bids.

For advertisers, PMPs offer meaningful advantages. Brand safety is higher because the publisher controls which advertisers participate. Inventory quality is typically better since PMPs are built around premium placements rather than remnant supply. The competitive environment is less crowded, which can improve win rates without requiring higher bids. Furthermore, first-party data agreements between buyer and publisher are common in PMP setups, enabling audience targeting not available in open exchange environments.

Header Bidding: How Publishers Maximize RTB Competition

Header bidding is a publisher-side technique that increases competition across RTB auctions by allowing multiple DSPs and exchanges to bid on the same impression simultaneously, before the publisher’s ad server makes a final decision. In the older waterfall model, publishers passed unsold inventory sequentially through a ranked list of networks. The first network to fill the impression won, regardless of whether a higher-paying buyer was available.

Header bidding solved this by running auctions in parallel. The publisher’s browser sends simultaneous bid requests to multiple exchanges through a JavaScript wrapper, collects all bids, and passes the highest to the ad server as the floor price for that impression. As a result, publishers earn more per impression because every qualified buyer competes at once rather than in sequence. For advertisers, header bidding means more competitive auctions on premium inventory, which makes accurate bid calibration even more important.

How RTB Auction Data Improves Campaign Performance

Every RTB auction generates data. DSPs that analyze auction logs systematically build a significant performance advantage over those that do not. Specifically, logs reveal which placements consistently clear at low prices, which audience segments attract heavy competition, and which times of day produce the best cost-per-conversion outcomes. This data also exposes floor price patterns that inform smarter bid floor strategies.

Well-optimized RTB campaigns improve over time because each auction cycle feeds back into the bidding model. A DSP tracking win prices, loss rates, and clearing price distributions continuously refines its valuation of each impression type. As a result, campaigns become more cost-efficient without reducing reach or conversion performance. This self-reinforcing optimization loop is one of the structural advantages of programmatic advertising over manual media buying, where optimization cycles take days rather than milliseconds.

Real Time Bidding for Web3 and Crypto Advertisers

For Web3 and crypto projects, RTB presents both an opportunity and a constraint. The opportunity is precision: RTB allows bidding only on impressions from users who match specific behavioral profiles, wallet-connected visitors, DeFi protocol users, or blockchain gaming communities. The constraint is infrastructure: mainstream ad exchanges restrict most crypto-related creatives regardless of compliance status.

Consequently, Web3 teams need access to RTB auction infrastructure built specifically for crypto inventory. AdsNetwork operates exactly this type of system, providing real-time bidding access across Web3-native publisher networks with floor prices, clearing mechanics, and audience data calibrated for crypto audiences. Because crypto audiences are concentrated on specific publisher types, win rate and floor price dynamics behave differently than on general exchanges. Understanding how RTB auctions clear in a crypto-native environment is therefore directly applicable to campaign cost efficiency for any Web3 project running paid acquisition.

Real Time Bidding: What You Now Know

Real time bidding is not a black box. Every clearing price, floor price threshold, and win rate signal is data. Teams that understand how RTB auctions work at the mechanics level can use that data to bid more accurately, reduce wasted spend, and improve campaign returns continuously. Teams that treat RTB as a set-and-forget channel leave significant efficiency on the table.

The most important takeaways: DSPs and SSPs operate as counterparts in every auction. Floor prices set the minimum clearing threshold. First-price auctions require bid shading to avoid overpaying. Win rate is the earliest diagnostic signal of bidding accuracy. And PMPs provide better inventory access with less competition than open exchange RTB. Ready to put this into practice? Explore AdsNetwork to run RTB-powered programmatic campaigns with crypto-native targeting.