The digital advertising ecosystem has undergone a fundamental transformation over the past decade, driven largely by the emergence of real-time bidding technology. What was once a relatively static marketplace for ad inventory has evolved into a sophisticated, lightning-fast auction system where billions of ad impressions are bought and sold in milliseconds. The real-time bidding market is currently valued at approximately $32 billion in 2025, with individual auctions completing in under 150 milliseconds. This technological revolution has democratised ad buying, enabled precise targeting, and created new efficiencies across the advertising value chain. Understanding how real-time bidding works has become essential for marketing technology professionals, because RTB now underpins the vast majority of programmatic advertising activity globally.
At its core, real-time bidding represents a fundamental shift in how digital advertising inventory is transacted. Rather than buying ad placements through direct negotiations or batch purchases, RTB enables advertisers to bid on individual ad impressions as they become available, in real time. The process begins when a user visits a website or opens an app. The publisher’s ad server immediately generates a bid request containing detailed information about that specific impression: the user’s location, browser type, device category, page content, previous behaviour, and sometimes even contextual signals about the surrounding content. This bid request is transmitted to an ad exchange, a neutral platform that facilitates the auction between multiple buyers.
The technical mechanics of RTB are extraordinarily complex, yet the speed at which they operate is what makes the technology revolutionary. When a bid request arrives at the ad exchange, it is simultaneously sent to multiple demand-side platforms, or DSPs. These are technology platforms that advertising agencies and advertisers use to programmatically purchase ad inventory across multiple publishers and ad networks. Each DSP receives the bid request and must make a split-second decision about whether to bid on that impression, and if so, how much to bid. This evaluation happens through sophisticated algorithms that assess the impression’s value to the advertiser based on factors including the user’s likelihood to convert, the fit with campaign objectives, current budget pacing, and competitive dynamics.
Understanding Auction Mechanics in Real-Time Bidding
Two primary auction models dominate the RTB landscape: first-price auctions and second-price auctions. In a first-price auction, the winning bidder pays exactly what they bid. This model creates an incentive for bidders to bid strategically lower than their true valuation, because overpaying for an impression represents wasted budget. Conversely, in a second-price auction, also known as a Vickrey auction, the winning bidder pays only the price of the second-highest bid plus a minimal increment. This model theoretically encourages bidders to bid their true valuation, because they will only pay what the next-highest bidder was willing to spend. For years, second-price auctions were the industry standard, but the shift toward first-price auctions began around 2018 and has now become predominant. Publishers adopted first-price auctions to capture more revenue, though this shift created new challenges for buyers in terms of bid optimisation.
Demand-side platforms and supply-side platforms are the two critical intermediaries that make RTB functional at scale. A DSP aggregates inventory from many publishers and ad exchanges, offering advertisers a unified platform to set targeting parameters, manage budgets, and optimise campaigns across hundreds or thousands of inventory sources. Supply-side platforms, or SSPs, serve the opposite function. They represent publishers’ inventory interests, connecting publishers to multiple ad exchanges and DSPs to maximise the revenue each publisher receives from their available ad space. The relationship between DSPs and SSPs is symbiotic: DSPs need access to quality inventory, while SSPs need access to buyers with sufficient budget and demand. This two-sided marketplace dynamic is what enables RTB to function efficiently.
Bid Shading and Supply Path Optimisation
In the context of first-price auctions, bid shading has emerged as a critical optimisation technique. Bid shading refers to the practice of bidding less than one’s true valuation for an impression, with the goal of achieving the best return on ad spend without overpaying. A bid shading algorithm might evaluate historical data about which impressions convert, at what price they typically clear, and what the user’s value is to the advertiser. Rather than bidding the maximum amount an advertiser is willing to pay, the algorithm bids strategically lower to improve margin whilst still winning valuable impressions. This has created an arms race in algorithm sophistication, with DSP providers continually refining their bid shading models to gain competitive advantage.
Supply path optimisation, or SPO, represents another layer of complexity in the RTB ecosystem. Because publishers’ inventory can be accessed through multiple pathways, direct connections to exchanges, through multiple SSPs, through header bidding wrappers, and through waterfall arrangements, the quality of impression data and the fees charged vary considerably by path. Supply path optimisation involves publishers and advertisers analysing these various paths to inventory and eliminating those that offer poor quality data, excessive latency, or excessive margin leakage. Publishers might discover that accessing inventory through one SSP includes significant data enrichment and attracts higher-quality buyers, whilst another path applies heavy compression and attracts lower-value bidders. SPO has become a significant focus area, as even small improvements in the paths that inventory travels can yield material improvements in revenue.
Floor Prices and Private Auctions
Publishers protecting their inventory quality and revenue have increasingly adopted floor prices in RTB auctions. A floor price is a minimum acceptable price that the publisher sets for ad impressions on their site. If no bidder exceeds the floor price, the impression remains unsold. Floor prices serve multiple strategic purposes: they prevent inventory from selling at excessively low prices, they signal to the market that a publisher values their inventory, and they encourage higher-quality bidding. Dynamic floor pricing, where the floor price adjusts based on real-time data about demand and impression characteristics, has become increasingly sophisticated. Some publishers now use machine learning models to set floor prices impression by impression, maximising expected revenue.
Private marketplaces, or PMPs, represent a hybrid between the open auction and direct-sold inventory. In a private marketplace, a publisher invites a select group of advertisers or agencies to bid on premium inventory under specific terms. The process still involves real-time bidding and auctions, but the participant pool is restricted and curated by the publisher. Advertisers value PMPs because they often offer better quality inventory, first look at premium placements, and reduced fraud risk compared to the completely open auction. Publishers benefit because PMPs allow them to command premium prices from their most desirable buyers whilst still leveraging the efficiency of real-time bidding technology.
Fraud and Latency Challenges
The speed and scale of RTB have unfortunately created opportunities for bad actors to perpetrate advertising fraud. Ad fraud in RTB environments typically takes several forms: invalid traffic from bots or non-human sources, domain spoofing where sites misrepresent their identity to command higher prices, viewability fraud where ads are purchased as if they will be seen by humans but actually load off-screen, and click fraud where automated systems generate fake clicks to waste advertiser budgets. The speed of RTB auctions, which complete in mere milliseconds, makes real-time fraud detection extraordinarily challenging. By the time fraudulent activity is detected and investigated, millions of fraudulent impressions may have already been transacted. Major advertisers and DSPs have invested heavily in fraud detection and prevention, but the problem remains persistent and evolving.
Latency represents another significant technical challenge in RTB environments. The milliseconds required for a bid request to travel from publisher to exchange to DSP and back to exchange to publisher can impact whether an ad actually appears before the page loads or the user leaves the site. High latency in the bidding process increases the likelihood that an impression opportunity will be missed. Publishers and technology providers have implemented various optimisations to reduce latency: geographically distributed servers, edge computing, pre-bidding where DSPs pre-fetch user data, and header bidding wrappers that run bidding logic directly within the publisher’s page. These technical innovations have become critical competitive differentiators in the RTB space.
Evolution Toward Programmatic Guaranteed
The progression of RTB technology continues to evolve, with increasingly sophisticated variants emerging. Programmatic guaranteed, also called programmatic direct, represents a convergence of the efficiency of programmatic buying with the certainty of direct ad purchases. In a programmatic guaranteed transaction, an advertiser and publisher agree on volume, price, and terms, but the transaction is executed programmatically rather than through manual negotiation. This retains the execution efficiency of RTB whilst reducing uncertainty about inventory availability. Preferred deals offer another variant, where advertisers receive first access to inventory at negotiated prices before it enters the open auction. These emerging models suggest that the future of digital advertising will involve a spectrum of buying mechanisms rather than pure open auction RTB.
| Auction Type | How It Works | Buyer Advantage | Seller Advantage |
|---|---|---|---|
| First-Price Auction | Winner pays their bid | Incentive for strategic bid shading and margin optimisation | Higher revenue per impression than second-price |
| Second-Price Auction | Winner pays second-highest bid plus one cent | Encourages truthful bidding by reducing overpayment risk | Lower revenue per impression but simpler buyer psychology |
| Programmatic Guaranteed | Pre-negotiated volume and price executed programmatically | Guaranteed inventory at agreed price with programmatic efficiency | Certainty of revenue and budget allocation |
| Private Marketplace | Invitation-only auction with curated buyers | Access to premium inventory and reduced fraud risk | Premium pricing from desirable advertisers |
| Component | Function | Key Considerations |
|---|---|---|
| Ad Exchange | Neutral marketplace that auctions impressions between buyers and sellers | Exchange fees, data transparency, latency impact on auction speed |
| Demand-Side Platform | Enables advertisers to bid on inventory and manage campaigns programmatically | Bid optimisation algorithms, audience targeting capabilities, integrations with CRM |
| Supply-Side Platform | Connects publisher inventory to multiple buyers and exchanges | Yield optimisation, data enrichment, fragmentation across multiple SSPs |
| Bid Shading Algorithm | Optimises bids based on expected conversion value and clearing prices | Model accuracy, competitive advantage, ethical considerations in strategic bidding |
| Header Bidding Wrapper | Runs auctions on publisher page before server-side request, reducing latency | Page load performance, partner integration complexity, reconciliation with server-side auctions |
The future of real-time bidding will continue to evolve as the industry addresses longstanding challenges and adapts to regulatory changes. Privacy regulations like GDPR and CCPA are fundamentally challenging the data-driven models upon which current RTB relies, pushing the industry toward first-party data strategies and contextual targeting. The deprecation of third-party cookies from web browsers will force significant changes to how audiences are identified and targeted in RTB environments. Simultaneously, improvements in fraud detection, latency optimisation, and bid shading algorithms will continue to refine the economics of RTB. For marketing technologists, staying current with RTB developments remains essential, because real-time bidding technology will continue to underpin programmatic advertising for the foreseeable future.
