Programmatic is still growing, but the rules of the game have changed. The IAB/PwC Internet Advertising Revenue Report says U.S. internet ad revenue reached $294.6 billion in 2025, up 13.9% year over year. Search remained the largest format at $114.2 billion, while video reached $78.0 billion and grew 25.4% year over year, the fastest pace among major formats. Bigger budgets usually sound fun, until you realize wasted impressions scale just as fast.
So, where does the next efficiency gain come from? Not from buying more inventory and not from stuffing another audience segment into a DSP and praying to the CPM gods. The stronger play is data-driven advertising built around curated supply: first-party data, contextual advertising, and intent data, all tied together before the auction starts.
Our public roadmap points in exactly that direction with planned AI-based audience curation, contextual targeting, and deeper auction analytics.
Table of contents
- What Is Data Curation in Programmatic Advertising?
- Why Traditional Targeting Is Breaking Down
- What Is Intent Data?
- What Is Contextual Targeting in 2026?
- Intent Data vs. Contextual Targeting: Which One Should You Use?
- Where Data Curation Fits In
- How Data Curation Improves Auction Performance
- The Future: AI-Based Audience Curation and Contextual Intelligence
- FAQ
What Is Data Curation in Programmatic Advertising?
IAB UK defines data curation as the intelligent selection, enrichment, and packaging of digital media inventory using trusted datasets and contextual signals. It can happen through prepackaged deals or real-time operations, which means it goes well beyond the old concept of a static PMP deal.
In plain terms, instead of buying a large pile of impressions and hoping the good ones are somewhere inside, you build smarter supply before the auction even starts.
That’s the key change happening right now. Curation is increasingly moving to the sell side, where supply-side curation improves scale, speed, precision, quality, and transparency. Real-time sell-side curation specifically turns static deal packaging into faster, impression-level decisioning.
Why does that matter? Because the supply side sees richer placement metadata and can act on it earlier than any DSP can. The pipe is smarter closer to the source.
Why Traditional Targeting Is Breaking Down
Behavioral targeting isn’t dead, it’s just less dependable than it used to be, and it’s getting less dependable faster.
Safari lists third-party cookie blocking as a default privacy feature, and Firefox blocks cross-site tracking cookies by default and stores them for the site that sets them. Moreso, Brave blocks trackers and cross-site cookies out of the box, and Chrome gives users and admins controls over third-party cookies, but cross-site tracking is already inconsistent across browsers. Yet, these are just a few examples.
On top of that, the UK ICO’s online tracking guidance is under review after legal changes in June 2025, so the policy picture is still maturing. And that’s not just in the UK. That’s the privacy side, and there’s also a pure efficiency problem that doesn’t get talked about enough.
Buy-side workflows face real signal loss and QPS (queries per second) constraints. Sell-side decisioning, on the other hand, can occur in around 10 milliseconds, compared with 100 milliseconds or more on the buy side. Targeting the source improves addressability, raises match rates, and cuts hops. You lose less data, faster, with fewer middlemen involved.
The old model is a privacy problem, sure, but it’s also an efficiency problem that’s always been hiding beneath the surface.
What Is Intent Data?
Intent data is the digital trail left by buyers as they research a category, capturing signals from content consumed, review sites visited, and topics they repeatedly research. The useful distinction is between first-party intent, which is signals collected on your own properties, and third-party intent, which is gathered across the broader web from research activity, publication reads, and review site visits.
The practical way to think about it: watch for spikes in online activity around topics tied to your product, then use those signals to target people while they’re actively in-market.
The performance argument for intent data is backed by real numbers. For instance, in a documented case, intent-driven programmatic campaigns helped cut cost per click by 80% and generate more receptive leads by focusing spend on people already showing buying signals. That said, as with any vendor case study, your own holdout tests should confirm whether the lift transfers to your specific context.
Intent data works best for B2B, fintech, high-consideration verticals, and any situation where timing matters more than reaching a broad audience that loosely fits your persona.
Where it gets unreliable is with weak or stale signals, missing quality thresholds, or treating any research activity as evidence of buying intent. Reading one article about enterprise software is meaningfully different from evaluating three vendors in the same week.
What Is Contextual Targeting in 2026?
Contextual targeting used to mean matching an ad for running shoes to a page that had the word “running” in the headline, and that’s a long way from what it means today.
The evolution of programmatic contextual targeting systems is one of the biggest shifts across the open web. Modern systems pre-screen pages using predictive science along with emotion and sentiment detection, with page-level classification across more than 200 vertical, seasonal, topical, and audience-proxy segments.
On the operational side, publishers pass contextual signals through SSP bid requests, and advertisers set rules so bids only fire when the content environment actually fits.
So what does that actually mean in practice? Contextual targeting in 2026 is the process of serving ads based on the full content environment a person is in at that moment, including the topic, the sentiment it evokes, the emotional state it triggers, and the scene or format being consumed. It’s not about who the user is in a database, it’s about what they’re reading, watching, or listening to right now, and what that moment makes them ready to receive.
The bigger story in 2026 is how contextual targeting has expanded into CTV and emotional alignment. Wurl’s BrandDiscovery product, for example, classifies streaming content scene by scene and scales across 95 billion CTV impressions per month. In one documented case, syncing a QSR ad with scenes that triggered hunger drove a 48% lift in sales and a 40% increase in store visits. Those aren’t marginal gains.
Research from Kantar adds useful context here: nearly 9 in 10 marketers believe creative testing needs to account for media context, and a third of strong TV ads lose their creative impact when moved to digital without contextual optimization. That’s a significant portion of creative value being left on the table just by ignoring the environment.
In 2026, contextual targeting means page, scene, sentiment, and moment. The topic alone is no longer enough.
Intent Data vs. Contextual Targeting: Which One Should You Use?
The short answer is that you don’t have to pick one. You do need to understand what each signal is actually answering, because they’re doing different jobs.
| Signal | Core Question | Best Use Case | Main Weakness |
|---|---|---|---|
| Intent data | Who is actively researching right now? | High-consideration buying journeys, ABM, B2B, fintech, any situation where timing matters more than broad reach | Can create false positives if the signal is weak, stale, or applied without quality thresholds |
| Contextual targeting | What is this person consuming right now, and what mindset does that create? | Privacy-safe reach, brand suitability, creative alignment, CTV, open-web campaigns where environment quality drives response | Good environment fit doesn’t prove purchase readiness on its own |
The cleaner frame: intent predicts demand, while context shapes receptivity. One tells you who might buy. The other tells you when the message is likely to land. The teams getting the most out of both aren’t treating them as competitors. They’re using them together.
Where Data Curation Fits In
Data curation is essentially the process by which those signals become tradable media.
Thus, curation pulls together publisher metadata, audience insights, contextual information, media quality signals, identity or ID-less identifiers, and publisher first-party data into real-time packages. Importantly, publishers and data providers can scale first-party data through curated audiences without relying on cookies or mobile IDs, and it works across browsers, apps, and OTT/CTV environments.
So if first-party data advertising is the asset, curation is the packaging layer. And if cookieless targeting is the goal, curated audiences combined with contextual signals are what make it tradable at scale. It’s not the most exciting framing, but the pipes are exactly where money quietly leaks or quietly compounds over time.
How Data Curation Improves Auction Performance
The strongest case for programmatic data curation comes down to auction math, and the numbers are hard to argue with.
Targeting closer to the supply improves addressability by reducing data leakage, increasing match rates, and reducing hops. Campaigns where data was applied via a sell-side curation layer saw 37% better CPM performance than those where the same data was applied on the buy side. That sounds counterintuitive at first, but it makes sense when you consider where the richer metadata actually lives. It’s not in the part of the pipe that gets the leftovers.
Publishers have a strong reason to pay attention here, too. The same study showed that curated deals drive 25%+ higher eCPMs, a daily incremental deal revenue uplift of up to 5% for many publishers and up to 10% for some, and up to a 14 percentage-point increase in spend from a broader set of buyers. Those aren’t rounding errors.
On the operational side, manual deal entry still causes mismatches, under-delivery, and confusion around ownership. Research from 2025 found that two-thirds of deals are configured with the desired supply yet deliver little or no revenue. That’s a startling number, and it suggests many programmatic “performance problems” are really workflow problems wearing a KPI costume. Standardizing deal sync and increasing transparency into seller, packager, and curator roles is where that gap gets fixed.
The Future: AI-Based Audience Curation and Contextual Intelligence
The realistic near-term future looks less like “AI replaces media buyers” and more like “AI makes curation, classification, and optimization fast enough to actually matter in real-time auctions.” IAB’s State of Data 2025 found that only 30% of brands, agencies, and publishers had fully integrated AI across the media campaign lifecycle. Half of those who hadn’t yet done so expected to by 2026.
At the same time, half the industry lacked a strategic roadmap, and nearly two-thirds named data quality, data protection, and tool fragmentation as the top barriers. That combination points clearly in one direction: your model is only as good as your signal infrastructure.
Sevio’s 2026 roadmap sits squarely inside that trend. The announced plans include AI-based audience curation and contextual targeting to turn publisher data, user behavior, and contextual signals into privacy-safe audience segments, alongside expanded auction visibility through Prebid analytics. The goal is better audience packaging, better effective CPMs, and clearer auction diagnostics with tighter feedback loops.
What makes that combination relevant isn’t just the technology. It’s the timing. Publishers are under pressure to activate first-party data more effectively, targeting signals are increasingly fragmented across browsers and environments, and the old model of behavioral targeting at scale is losing reliability fast. Sevio’s roadmap addresses all three of those problems in a single, integrated direction: smarter sell-side curation powered by AI, with the auction transparency to measure what it’s actually doing.
Based on where major sell-side platforms and industry standards bodies are pointing, the practical playbook for programmatic teams isn’t difficult: start with one vertical, package consented first-party and contextual signals into curated deals, layer intent where you have solid buying-cycle evidence, then benchmark outcomes against a non-curated control. No big bang required. Just cleaner infrastructure and honest measurement.
FAQ
It’s becoming the baseline, not a workaround. Contextual targeting, first-party data advertising, and curated audience packages all work without third-party cookies and have shown real performance gains in documented campaigns. With cross-site tracking already broken by default across Safari, Firefox, and Brave, cookieless approaches aren’t the backup plan anymore.
First-party data is what you collect directly from your own audience through site behavior, email engagement, or subscriptions. Third-party intent data is gathered across the broader web by tracking research activity on publications, review sites, and content networks. The first tells you what your known audience does; the second tells you who outside that audience is actively researching a purchase right now.
A PMP is a specific deal format; data curation is what makes that deal smarter. Curation layers audience insights, contextual signals, and quality filters onto inventory before it enters a deal package, so the resulting PMP is built around signal rather than just supply access. Without curation, a PMP is essentially a reserved pipe. With it, it’s a targeted one.
Use intent data when timing is the priority: B2B, high-consideration purchases, or any campaign where catching someone mid-research matters more than broad reach. Use contextual targeting when environment quality, brand suitability, or creative alignment are the priorities, especially for CTV, open web, and privacy-sensitive campaigns. In practice, the strongest campaigns use both together rather than treating them as alternatives.
It works for B2B, just differently. The strongest use case is brand suitability and environmental quality: appearing in credible editorial environments where a professional audience consumes relevant content. Layering contextual signals with intent data is especially effective here, since it combines “who is researching this topic” with “where they’re reading about it.”
It means building addressable audiences through signals that don’t rely on third-party cookies: consented first-party data, page and scene-level contextual signals, publisher-side packages from logged-in user data, and ID-less solutions like cohort-based targeting. The core shift is to target based on what someone is doing right now and the environment they’re in, rather than following them across sites.
Final Thoughts
Data curation complements intent data and contextual advertising, not replaces them, and it’s the mechanism that makes both more usable, more transparent, and more auction-efficient.
Intent data helps you identify demand, whereby contextual targeting helps you match the message to the right moment. Curation is what turns both into something buyable at scale across a market that no longer offers easy, universal tracking.
For programmatic teams, the move forward isn’t buying more data. It’s buying fewer, better chances to win, through supply-side packaging, consented first-party data, semantic context, and tighter measurement.
The direction from Sevio’s roadmap, IAB standards work, and sell-side platform research is consistent: quality-first, privacy-safe curation is where better programmatic results are heading. Vendor numbers should still be validated with your own holdouts, but the direction itself is pretty hard to argue with at this point.
Was this helpful?
