How We Got Here

Programmatic media did not begin as a cautionary tale. It emerged from a genuine effort to solve real problems of scale, coordination, and efficiency in digital advertising, driven by optimism about what automation and data could enable for advertisers and publishers alike.

Understanding how those early solutions shaped today’s ecosystem is essential for anyone interested in restoring and strengthening economic confidence across the programmatic supply chain.

The foundations of today’s programmatic system trace back to the late 1990s and early 2000s, as the open web expanded rapidly. Thousands of new publishers entered the market, most without the scale, sales infrastructure, or advertiser relationships required to sell inventory directly. At the same time, advertisers faced growing difficulty buying digital media efficiently across a fragmented landscape.

Early ad networks emerged as a pragmatic response. They aggregated inventory from many publishers, particularly smaller or long tail sites, and bundled it into scalable buying opportunities. For advertisers, this simplified access to reach. For publishers, it created demand for inventory that might otherwise have gone unsold.

These networks created real value by applying data, optimization, and pricing strategies that individual buyers and sellers could not easily deploy on their own. At the same time, they embedded intermediation and margin capture into the system from the outset.

Perhaps most importantly, ad networks normalized a durable idea that persists today: that inventory could be abstracted, packaged, and resold, with economic outcomes determined somewhere between buyer and seller rather than explicitly negotiated by either.

By the mid 2000s, the limits of static network buying became more apparent. This period saw the emergence of ad exchanges and real time bidding, which enabled individual impressions to be auctioned dynamically rather than sold in pre packaged blocks.

RTB promised greater efficiency and improved price discovery. Advertisers could bid on impressions as they appeared. Publishers could expose inventory to broader demand while retaining more pricing control. Adoption accelerated through the late 2000s and early 2010s, and RTB became the backbone of modern programmatic media.

Crucially, exchanges did not replace ad networks so much as extend their logic. Aggregation and abstraction remained, but at far greater speed and scale. Automation increased, and so did the number of intermediaries participating in each transaction.

Historically, digital media buying was anchored in location. Advertisers knew where their ads ran, specific publishers, sections, or environments, and evaluated value accordingly. Even when buying was imperfect, the unit of analysis was legible.

As programmatic buying matured, that anchor shifted. Decisions increasingly occurred at the level of the platform rather than the place. Advertisers optimized within DSPs, audiences, and algorithms, often without consistent or granular visibility into where impressions ultimately delivered, particularly across the open web.

Location became an output rather than an input.

For advertisers, this reduced operational friction but weakened economic line of sight. Confidence increasingly rested on platform reported performance rather than on understanding how spend was distributed across publishers. For publishers, inventory became more abstract, treated as interchangeable supply rather than differentiated environments with intrinsic value.

Alongside this shift, the industry moved from contextual buying toward audience addressability. Cookies enabled impressions to be valued based on inferred user attributes rather than content alone.

Contextual buying was always understood to involve uncertainty. Ads were placed based on environments assumed to be relevant, not verified individual identity. That uncertainty was visible and accepted because the inputs, content and placement, were legible.

Audience buying promised to reduce that uncertainty. It appeared to offer more precision, less waste, and greater accountability. The optimism was real and well intentioned.

What was less acknowledged was that audience buying did not eliminate uncertainty. It relocated it.

Cookie based buying introduced a less visible margin of error. Identifiers degraded over time as users cleared cookies, switched devices, or were represented probabilistically. Match rates varied widely by environment and format.

To sustain scale, buying systems increasingly relied on modeled identity, look alike audiences, and expansion techniques. These approaches worked operationally, but they diluted original audience definitions.

Performance often remained acceptable in aggregate, masking variability in match quality and signal fidelity. Uncertainty became embedded in models and assumptions rather than environments, making it harder to observe and harder to challenge.

As this infrastructure matured, the meaning of a CPM changed.

In the early years of digital advertising, a CPM largely reflected the price of media inventory in a specific environment. Over time, CPMs evolved into bundled economic signals that included inventory, technology costs, data and audience fees, operational labor, and intermediary margin.

For advertisers, this simplified buying but complicated interpretation. Two impressions at the same CPM could reflect very different underlying economics. For publishers, higher buyer side CPMs did not necessarily translate into higher yield, as value was increasingly captured upstream.

Once CPMs became abstractions rather than media prices, isolating where value was created and where it was extracted became more difficult for both sides.

This challenge was compounded by structural constraints within the programmatic ecosystem. Platform contracts often limit the ability of advertisers and publishers to share detailed spend and revenue information directly with one another.

Advertisers see CPMs and performance metrics without insight into publisher yield. Publishers see revenue without visibility into buyer side pricing. Even when both parties want to understand outcomes more clearly, they are often forced to rely on platform mediated reporting rather than shared economic evidence.

The parties with the greatest interest in understanding programmatic economics are frequently the least able to compare them directly.

As programmatic spend scaled through the 2010s, agency holding companies adapted by building internal trading desks and principal based buying entities.

Acting as principals allowed these entities to recognize the full amount of client media spend as top line revenue rather than only a service fee. Over time, incentives, reporting structures, and procurement dynamics increasingly favored internal buying operations.

What began as an exception gradually became a standard operating pattern across parts of the buy side supply chain.

Industry research, including multiple ANA commissioned studies, increased awareness of programmatic supply chain complexity and economic variability. However, incentives remained largely unchanged.

Performance metrics continued to dominate decision making. Aggregation remained rewarded. Publishers remained dependent on intermediated demand to access budgets at scale.

The programmatic ecosystem that exists today is the cumulative result of these choices. It is technologically sophisticated and operationally efficient, yet often difficult for advertisers and publishers to engage with confidently from an economic standpoint, particularly on the open web where delivery locations are frequently obscured.

This outcome was not driven by bad intent. It was driven by incentives that consistently rewarded scale, abstraction, and margin over clarity and directness.

Understanding how those incentives took hold, and why they persist, is a prerequisite to any realistic effort to improve outcomes for the buyers and sellers the system exists to serveHow We Got Here

Programmatic media did not begin as a cautionary tale. It emerged from a genuine effort to solve real problems of scale, coordination, and efficiency in digital advertising, and from widespread optimism about what automation and data could enable for advertisers and publishers.

Understanding how those early solutions shaped today’s ecosystem is essential for anyone interested in improving economic confidence across the programmatic supply chain.