Programmatic Advertising Fundamentals · RTB, DSPs, SSPs & the Ecosystem

Programmatic advertising is the automated buying and selling of digital advertising inventory using software, algorithms, and real-time auctions. Instead of a media buyer negotiating with a publisher's sales team, agreeing a rate card price, and signing an insertion order — a process that could take days or weeks — programmatic executes the equivalent transaction in approximately 100 milliseconds, every time a user loads a webpage.

The Interactive Advertising Bureau (IAB), the industry body that sets programmatic standards, defines programmatic as "the automation of the buying, placement, and optimisation of media inventory via a bidding system." The three key words are: automation (no human deal-making required for each transaction), placement (the ad is served to a specific impression in a specific context), and bidding system (price is determined by competition in real time, not by a pre-agreed rate card).

The scale of programmatic is significant. According to IAB and eMarketer documented estimates, programmatic now accounts for more than 90% of US digital display advertising spending — meaning that almost every banner ad, in-stream video ad, or connected TV pre-roll you see has been bought and sold through an automated auction in the fraction of a second before it appeared on your screen.

Programmatic matters to every digital marketer — not just those who run display campaigns — because it underpins the delivery infrastructure of display, video, audio, connected TV, and increasingly digital out-of-home advertising. Understanding how it works enables better briefing of media agencies, better interpretation of campaign data, better fraud protection decisions, and more informed budget allocation across channels.

Real-Time Bidding (RTB) is the most common programmatic transaction type. It is a fully automated auction that occurs every time a user loads a page that has programmatic ad inventory. The full sequence takes place in the time between a user clicking a link and the page appearing in their browser.

Here is the exact sequence:

1. User loads a page. A user navigates to a webpage that has programmatic display inventory — for example, a news article with a banner ad slot.
2. Publisher's ad server fires a bid request. The publisher's Supply-Side Platform (SSP) sends a bid request to one or more ad exchanges. The bid request contains: the ad slot dimensions and position; the page URL and content category; available user data signals (if consent exists) including audience segments, device type, browser, geographic location, and time of day; and a timestamp.
3. Ad exchange distributes the bid request. The ad exchange distributes the bid request simultaneously to all connected Demand-Side Platforms (DSPs). This happens in parallel, not sequentially.
4. DSPs evaluate and bid. Each DSP receives the bid request and runs it through its bidding logic in milliseconds. The DSP checks: does this impression match any active campaign's targeting criteria? If yes, what is the maximum bid for this impression based on the campaign's bidding strategy? DSPs that want to compete submit their bids back to the ad exchange.
5. Auction resolution. The ad exchange collects all bids and determines the winner. In a standard second-price auction (the Vickrey model, documented in academic auction theory since 1961), the winning bidder pays the second-highest bid plus one cent — not their own maximum bid. This creates an incentive for honest bidding at true maximum value. In first-price auctions (increasingly common after the industry moved away from second-price in 2019), the winner pays their own bid.
6. Ad is served. The winning DSP's ad server delivers the creative to the publisher's page. The entire process — bid request, auction, winner determination, ad delivery — completes in 50–100 milliseconds.
7. Impression is recorded. The ad server logs the impression. Verification vendors (if integrated) measure viewability and check for invalid traffic. Attribution systems record the impression for later conversion attribution.

The programmatic supply chain involves multiple technology layers, each serving a specific function. Understanding every layer is important for understanding where costs, fraud, and inefficiency enter the chain.

The cost of the supply chain — the percentage of each advertiser pound that goes to technology fees rather than reaching actual users — is called the "tax" on programmatic. The ANA and ISBA have both published documented research showing that in open programmatic, a significant percentage of advertiser spend is absorbed by the technology layers before reaching publisher revenue. This is one reason why private marketplace deals and direct programmatic arrangements often show better effective CPMs for publishers despite lower gross CPMs for advertisers.

Not all programmatic transactions are open RTB auctions. The industry has developed a spectrum of deal types that offer different levels of inventory exclusivity, price certainty, and control — allowing advertisers to access premium inventory while still using programmatic infrastructure for delivery and measurement.

The choice between deal types is primarily a trade-off between cost efficiency and inventory quality/certainty. Open auction provides the lowest CPMs but the least control over where ads appear. Programmatic guaranteed provides the highest certainty but at the cost of negotiation time and typically higher CPMs. Most sophisticated programmatic strategies use a combination: programmatic guaranteed for high-value brand placements where certainty matters, PMPs for premium contextual environments, and open auction for scale and efficiency.

Before programmatic, all display advertising was bought directly: a media buyer called or emailed a publisher's sales team, negotiated a CPM rate for a specific placement over a specific time period, signed an insertion order, and sent creative. This process typically took days to weeks and required a personal relationship with each publisher's sales team.

Most large advertisers today use both models, allocating them by objective: programmatic for audience targeting at scale and for campaigns where real-time optimisation matters; direct deals for high-impact premium placements (homepage takeovers, sponsorship integrations, contextually exclusive positions) where guaranteed premium placement justifies the negotiation overhead and higher CPMs.

The origins of programmatic lie in the problem of unsold display inventory. In the late 2000s, publishers were running large amounts of display inventory that their direct sales teams could not sell — remnant inventory that was going unused. Ad networks emerged to aggregate this inventory and sell it in bulk at low CPMs, but they lacked transparency and precision.

DoubleClick, acquired by Google in 2008, launched one of the first real-time bidding exchanges in 2009. The critical innovation was moving from static CPM deals to per-impression auctions — enabling buyers to value each impression individually based on audience data rather than paying a flat rate for a generic placement. This shift transformed the economics of display advertising: instead of paying for 100,000 impressions to reach a broad demographic, an advertiser could pay premium rates only for impressions served to users who matched specific audience criteria.

The subsequent decade saw rapid expansion: SSPs emerged to help publishers manage competing programmatic demand; DMPs emerged to provide audience data; verification vendors emerged to measure viewability and detect fraud; and programmatic expanded from display to video (2013), mobile (2014), audio (2017), connected TV (2018), and digital out-of-home (2019–2020).

The most significant structural challenge to programmatic has been the deprecation of third-party cookies — the tracking mechanism that enabled cross-site audience targeting. Safari blocked third-party cookies in 2017; Firefox followed in 2019; Chrome's phaseout, initially scheduled for 2022, was delayed multiple times before Google committed to Privacy Sandbox as a cookieless alternative. This transition is reshaping how audience targeting works in programmatic throughout 2024–2026.

Programmatic infrastructure now powers advertising delivery across multiple media channels beyond its original display origins:

Display (web and mobile). The original programmatic channel. Banner ads, native ads, and rich media placements across websites and apps. Standard IAB formats are defined in the IAB New Ad Portfolio specification. Still the highest-volume programmatic channel by impression count.

Programmatic video. In-stream (pre-roll, mid-roll, post-roll), out-stream (plays in content feed), and in-banner video. Priced at significantly higher CPMs than display due to higher engagement. The IAB VAST and VPAID standards govern video ad serving. YouTube's programmatic buying is managed separately through Google Ads and DV360 rather than through the open exchange.

Programmatic audio. Podcast and streaming audio advertising (Spotify, Pandora, iHeartRadio) bought programmatically. The IAB DAAST standard governs digital audio ad serving. Growing rapidly as podcast consumption increases but with different targeting and measurement challenges due to cookieless audio environments.

Connected TV (CTV). Advertising on smart TVs, streaming sticks (Roku, Amazon Fire), and gaming consoles reaching streaming content viewers. CTV programmatic is purchased through specialised DSPs and directly through streaming services. Measured by completion rate and brand lift rather than click-through rate. Detailed in the Connected TV guide.

Digital Out-of-Home (DOOH). Programmatic buying of digital billboard, retail screen, and transit screen advertising. Triggered by real-world conditions (weather, traffic, time of day) and audience data from mobile devices near the screen. An emerging channel that brings programmatic efficiency to physical world advertising formats.

Understanding programmatic economics means understanding where money flows through the supply chain — and where it leaks. The documented concern about programmatic transparency centres on how much of each advertiser pound actually reaches publisher revenue versus being absorbed by technology fees.

The ISBA Programmatic Supply Chain Transparency Study (2020 and 2023 updates) — one of the most rigorous documented studies of programmatic economics — found that in a sample of major UK advertiser campaigns, only 51p of every £1 of advertiser spend reached publishers in 2020, with 15p unattributable to any identified party in the supply chain. The 2023 follow-up showed improvement but significant fees remained. These findings drove increased adoption of supply path optimisation (SPO) — the practice of identifying and preferring direct, lower-cost paths to quality inventory.

Effective programmatic buyers manage these costs through: direct deals (bypassing exchange fees); supply path optimisation (reducing the number of intermediaries between DSP and publisher); and negotiated DSP fee structures (larger budgets command lower percentages). For budget allocation purposes, the working assumption should be that 25–40% of open-auction programmatic spend goes to technology fees rather than delivering ads to real users.

Sources & Further Reading