Programmatic advertising has just crossed a landmark threshold. Global programmatic ad spend has surpassed $200 billion in 2026, cementing its position not as a media tactic but as the core infrastructure through which the digital advertising industry operates. Nearly every digital display ad you see — on a website, a streaming platform, a connected TV, or a mobile app — was placed through a programmatic system.
But as programmatic has scaled, a problem has grown quietly alongside it. The same open exchange that gave advertisers access to billions of impressions across millions of sites has also become a breeding ground for waste, fraud, and brand safety risk. In 2025, programmatic inefficiencies and wasted spend were estimated to total approximately $26.8 billion globally. That is not a rounding error. That is a structural problem.
In 2026, the industry's response is clear: the era of spray and pray is over. The most sophisticated advertisers are moving away from the open exchange and toward curated inventory, private marketplaces, and AI-driven quality controls. This shift is the defining story of programmatic advertising right now — and understanding it is essential for any business investing in digital media.
$200B+
global programmatic spend in 2026
58%
of buyers increasing programmatic budgets this year
$26.8B
estimated wasted on low-quality inventory in 2025
What Is the Open Exchange — and What Went Wrong?
The open exchange is a real-time auction environment where publishers make their unsold ad inventory available to any buyer. A DSP (Demand-Side Platform) bids on available impressions in milliseconds based on audience data and targeting criteria. The highest bid wins the placement, and the ad is served. In theory: efficient, scalable, cost-effective. In practice: complicated.
Over the past several years, three compounding problems eroded the open exchange's value proposition:
1. Ad Fraud at Industrial Scale
Bad actors discovered that the open exchange's scale and automation created enormous opportunities for fraud. Bot traffic, domain spoofing, and invalid impressions flooded the supply chain. Despite improvements in fraud detection, the problem persisted because the economics of the open exchange — where any inventory could be made available with limited verification — made quality control structurally difficult.
2. The AI Content Explosion
The proliferation of AI-generated content websites created a new category of low-quality inventory. These sites produce content at scale, attract marginal traffic, and monetise through programmatic ads. The result: an advertiser's campaign might technically reach a human audience, but in a context that generates zero meaningful engagement and carries real brand safety risk. Research found that 54% of advertisers believe generative AI has contributed to a decline in overall media quality across the open exchange.
3. The Measurement Collapse
As zero-click search and AI Overviews changed how users navigate the web, the metrics that justified open exchange spend started breaking down. Click-through rates fell. Attribution became murkier. Advertisers paying open exchange CPMs were increasingly unable to connect those impressions to any meaningful business outcome. Volume was being optimised. Value was not.
The Core Problem
You can reach a billion impressions through the open exchange. But if 30% are fraudulent, 40% are on low-quality AI-generated content, and the remaining 30% are on sites where your audience is not genuinely engaged — you have not bought media. You have bought numbers.
The Rise of Curated Inventory and Private Marketplaces
The industry's response has been decisive. Curated inventory and Private Marketplace (PMP) deals have moved from a premium option used by large brand advertisers to the dominant framework for serious programmatic media buyers. According to ANA's Q2 2025 Programmatic Transparency Benchmark, curated marketplaces and PMP transactions now represent the majority of programmatic spend. This is not a gradual trend — it is a structural shift.
Private Marketplaces (PMPs)
A PMP is an invite-only auction environment where a publisher offers inventory to a select group of approved buyers at negotiated terms. Because both parties have entered a direct relationship, the inventory is verified, the context is known, and brand safety is significantly easier to guarantee. PMPs currently account for approximately 30% of programmatic transactions and are growing consistently quarter on quarter.
Programmatic Guaranteed
A step further than PMPs, Programmatic Guaranteed deals involve reserved inventory at fixed pricing — essentially combining the automation and audience targeting of programmatic with the certainty of a direct buy. An advertiser knows exactly where their ads will appear, at what CPM, and in what volume. There is no auction uncertainty. This model is particularly valuable for brand-sensitive campaigns and for inventory on premium publishers that cannot afford to let their best placements go through the open exchange.
Curated Packages
Curation sits between the open exchange and full PMPs. DSPs and SSPs package open exchange inventory that meets defined quality criteria — verified human audiences, contextually relevant environments, pre-bid fraud filtration. The buyer gets the scale of the open exchange with meaningful quality controls layered on top. Research shows that 41% of advertisers now cite curated deals as a primary path to higher campaign ROI.
Roevex Perspective
When we build programmatic campaigns, we treat inventory quality as a first-order variable — not an afterthought. Every dollar spent in a verified, brand-safe environment performs differently from a dollar spent blindly in the open exchange. The CPM might be higher in a PMP. The cost per genuine business outcome is almost always lower.
Connected TV: The Highest-Quality Inventory in Programmatic Right Now
No area of programmatic is moving faster or attracting more investment than Connected TV (CTV). In 2026, nine streaming platforms are projected to individually exceed $1 billion in ad revenue. CTV now accounts for 20% of US media consumption. And CTV is the fastest-growing segment of programmatic buying — for good reason.
As Netflix, Amazon, Disney+, and others launched ad-supported tiers, they created an enormous surge of premium, verified inventory: audiences that are authenticated, engaged, and consuming content in a high-attention environment. This is precisely the kind of inventory that programmatic buyers have been searching for.
- Verified human audiences: CTV viewers are authenticated users. The fraud problem that plagues display advertising is structurally much harder to execute in streaming environments.
- High-attention context: A 30-second non-skippable ad in the middle of a streaming drama commands a fundamentally different quality of attention than a banner ad on a cluttered webpage.
- Cross-device attribution: CTV campaigns can now be connected to subsequent behaviour on mobile and desktop, enabling full-funnel measurement of streaming ad impact.
- Incremental reach beyond linear TV: CTV audiences are distinct from traditional broadcast viewers — younger, more streaming-native, and increasingly unreachable through conventional TV buying.
Budget is following this shift decisively. Research from Comscore found that CTV is expected to see the largest increase in programmatic budget share of any channel in 2026, with 14% of that CTV spend being entirely net new investment — not reallocated from any other channel.
Emerging Format
In February 2026, Samsung Smart TVs introduced new homescreen display and video ad formats that appear before content starts — turning the TV interface itself into premium programmatic inventory. This is the direction CTV is heading: high-visibility, premium placements in authenticated, brand-safe environments that the open exchange cannot replicate.
How AI Is Changing the Way Programmatic Campaigns Actually Run
AI has been part of programmatic since its early days, powering the real-time bidding algorithms that evaluate and price impressions in milliseconds. What has changed in 2026 is the depth and scope of AI's role. The industry is moving from AI as an optimisation tool to AI as the primary decision-maker across the entire campaign lifecycle.
Pre-Launch Intelligence
Historically, programmatic followed a launch-then-optimise cycle: a campaign would go live, accumulate data, and then be adjusted based on results — a process that took days or weeks. In 2026, leading media buyers are front-loading intelligence. AI models are used before launch to forecast performance, validate creative against target audiences, and identify supply paths likely to deliver quality at scale. Research found that 31% of advertisers want AI predictive models to forecast performance before campaigns go live, and 40% want to pre-test creative with synthetic audiences.
Agentic Buying
The most significant development emerging in programmatic is agentic AI — systems that do not just optimise campaigns but actively negotiate media placements, adjust deal structures, and manage supply paths without human intervention on each individual decision. Industry leaders describe this as fundamental: humans define strategic intent and set guardrails, AI handles activation. Early agentic buying systems are already operating in parts of the programmatic ecosystem, and their impact will be profound as they scale.
Real-Time Quality Monitoring
One of the most immediate applications of AI in 2026 is real-time inventory quality monitoring. Pre-bid fraud filtration, contextual intelligence tools, and AI-driven delivery analysis allow buyers to identify low-quality environments before an impression is purchased, rather than discovering the problem in a post-campaign audit. The practical impact: less wasted spend, cleaner optimisation signals, and campaign decisions based on genuine engagement rather than inflated volume.
The Paradox of AI in Programmatic
As AI generates more content across the web, it is also generating more fraudulent and low-quality programmatic inventory. The same technology making campaigns smarter is making the supply chain noisier. The solution is not to avoid AI in media buying — it is to use AI specifically to filter out what AI is doing to media quality. It requires fighting fire with fire.
What Smart Programmatic Strategy Looks Like in 2026
Given everything above, what does a well-structured programmatic approach look like for an advertiser right now? Here is the framework Roevex applies when building and managing programmatic campaigns.
Start with Signal Quality, Not Reach
The first question is not 'how many impressions can we buy?' It is 'what data signals do we have, and how clean are they?' First-party data — CRM records, email lists, site behavioural data — is the foundation of every high-performing programmatic campaign. It is also the most privacy-resilient signal as the industry continues to move away from third-party cookie dependence. Fragmented or unstructured first-party data is one of the most common reasons programmatic campaigns underperform, and it is fixable before a single impression is purchased.
Choose Your Inventory Environment Deliberately
Map your campaign objectives to inventory environments before you start buying. Brand awareness campaigns targeting a premium audience require different inventory than a direct-response performance campaign. Define your supply path — open exchange, curated packages, PMPs, or programmatic guaranteed — based on what outcome you are optimising for, not on what is cheapest. The lowest CPM is almost never the most efficient path to a business outcome.
Build CTV Into Your Mix
If you are not allocating a meaningful portion of your programmatic budget to CTV in 2026, you are missing the highest-attention, lowest-fraud inventory environment in digital advertising. Start with a test allocation, measure full-funnel impact carefully, and scale what works. CTV does not replace digital display — it anchors and amplifies the rest of your mix.
Resurrect Contextual Targeting
With audience-based targeting becoming more complex in a privacy-first world, contextual intelligence is making a significant comeback. Serving ads in content environments directly relevant to your product — rather than solely chasing a user profile — delivers stronger engagement and requires no personal data. Modern contextual tools analyse page-level content, sentiment, and category relevance in real time. This is not the blunt instrument it was five years ago.
Define Quality Metrics Before Launch
Decide what quality means for your campaign before the campaign goes live. Attention metrics, viewability thresholds, verified human impression rates, cost per verified outcome rather than cost per click — having these defined upfront means every optimisation decision is made against genuine value, not proxy metrics that look good in a report but do not connect to commercial results.
Final Thoughts
Programmatic advertising in 2026 is not the same game it was three years ago. The open exchange still exists and still has a role, but the era of buying impressions at scale with minimal regard for quality is ending — driven by advertiser frustration, AI-generated inventory pollution, and the availability of meaningfully better alternatives.
The shift toward curated inventory, private marketplaces, CTV, and AI-driven quality controls is not making programmatic more complicated for the sake of it. It is making programmatic deliver what it always promised: the right ad, in the right environment, reaching a real human who is actually likely to care.
That is harder to achieve than buying a billion cheap impressions. It is also the only thing that actually works.
Running programmatic campaigns that are not delivering?
Roevex audits programmatic setups and rebuilds them around quality, not volume. Get in touch.