Programmatic advertising did not start as the engine of digital media. It started as a clearance sale. In its early years, automated buying was where publishers sent the inventory they could not sell directly: the unsold banners, the late-night impressions, the pages nobody wanted to pay a premium for. Real-time bidding was, for a long time, almost a synonym for remnant.

That is no longer true. Programmatic now carries most of the money in digital display and video, and connected TV is the fastest-growing part of it. To understand where it is going, it helps to look at the three phases it has already lived through.

Phase one: automating the leftovers (2008 to 2014)

The first ad exchanges and real-time bidding systems appeared at the end of the 2000s. The idea was simple and powerful: instead of negotiating a campaign in advance, a buyer could look at a single impression, decide what it was worth, and bid on it in the few hundred milliseconds it takes a page to load.

The industry quickly needed a common language for this. That language became OpenRTB, the protocol that still defines how a bid request and a bid response are structured today. Almost every SSP, DSP and ad server speaks some version of it.

In this phase, the value proposition was efficiency. Buyers got scale and targeting. Publishers got a way to monetize inventory that would otherwise go unsold. Nobody was talking about premium inventory yet.

Phase two: premium goes programmatic (2014 to 2020)

Around the middle of the 2010s, buyers started asking for something new: the ability to buy the best placements programmatically, not just the leftovers. Two changes made it possible.

The first was header bidding. By 2015, publishers were letting several exchanges bid on the same impression at the same time, before their ad server made a decision. It broke the old "waterfall", where demand partners were called one after another, and it pushed the market toward first-price auctions. In March 2019, Google announced it would move Google Ad Manager to a first-price auction, which effectively made first price the industry standard.

The second change was transparency. As money flowed in, so did fraud and domain spoofing. The IAB Tech Lab answered with a set of simple but important standards:

  • ads.txt (2017), a file where publishers declare who is authorized to sell their inventory, followed by app-ads.txt for apps and CTV.
  • sellers.json and the SupplyChain object (finalized in July 2019), which let buyers see every company that touched a bid request on its way from the publisher.

The same year, global programmatic spending passed 100 billion dollars for the first time. Programmatic had become the default way to buy digital media.

Phase three: CTV, privacy and consolidation (2020 to today)

The current phase has three big themes.

Connected TV. Streaming has turned the television screen into a programmatic environment. According to the IAB's 2026 Outlook study, US CTV ad spend is expected to grow 13.8% in 2026, faster than the overall ad market. CTV brings premium screens and high CPMs, but it also brings new fraud risks, because a TV app does not have a browser and is harder to verify than a web page.

Privacy and identity. For years the industry prepared for the end of third-party cookies in Chrome. In 2025, Google confirmed it would keep third-party cookies and later retired its Privacy Sandbox project. The result is not a return to the old world. Regulation, consent requirements and walled gardens still limit what data can travel in a bid request, and the market keeps moving toward contextual signals, first-party data and better supply information.

Consolidation and scrutiny. In April 2025, a US federal court found that Google had illegally monopolized the markets for publisher ad servers and ad exchanges. Whatever the final remedies, the case has put the economics of the supply chain, take rates, auction rules and transparency, under a spotlight that is not going away.

What the present looks like

In 2026, programmatic is mature, but it is not simple. A typical impression can pass through a publisher, an SSP, sometimes a reseller, an exchange, a DSP and a verification vendor before an ad is shown. Each step adds cost and latency, and each step is a place where value can leak.

The IAB expects US ad spend to grow 9.5% in 2026. Money is not the problem. The problem is efficiency: buyers want proof that their budget reaches real people in real environments, and publishers want a fair share of what buyers pay.

Where it is heading

Three directions are already visible.

  1. Fewer, better paths. Buyers increasingly favor short, transparent supply paths: direct relationships, a clean sellers.json entry and a complete supply chain object. Inventory that cannot prove where it comes from is losing value.
  2. Verification before the bid, not after. Checking an impression after it has been bought is too late. Quality checks are moving upstream, into the moment before a request is even sent to a buyer.
  3. Prediction instead of broadcasting. The classic model sends every request to every buyer and lets the auction sort it out. That wastes bandwidth, money and compute on both sides. The next step is to predict which buyer is most likely to want a given request and route it there.

That last point is the reason we built PeopleAds the way we did. Our platform verifies every request (ads.txt, app-ads.txt, sellers.json and the supply chain object), scores it with a prediction model, and routes it to the buyers most likely to bid. We send less, and buyers spend more on what they receive.

Programmatic started by selling what nobody wanted. Its next chapter is about sending each impression only where it is wanted.

Sources