Performance is being redefined across the open internet. Advertisers are no longer just buying channels and hoping outcomes follow. They’re demanding systems that take responsibility for results, dynamically move spend, and clearly explain how performance is achieved. In short, marketers want accountable automation rather than manual optimizations, and they expect every dollar to work harder and smarter toward real business outcomes.
Pinterest’s recent acquisition of tvScientific is one signal of this shift. The move combines Pinterest’s intent-rich user data with tvScientific’s ability to extend targeting, activation, and full-funnel measurement beyond a walled garden and into the open web. tvScientific was built around a simple idea: TV should behave more like performance media – outcomes over impressions, optimization toward business results, and measurement that feels closer to search or social than old-school GRPs. Pinterest brings the missing ingredient: intent. People plan on Pinterest – they save, they signal future behavior. The bet is that combining intent-rich data with a performance-native execution layer makes TV advertising accountable in a way it’s never been before. That bet matters because it reflects a broader shift already underway: Advertisers are reallocating budgets toward systems that can reliably turn exposure into outcomes. And that shift does not stop at CTV.
Performance Across Channels
CTV became the starting point for the performance conversation because it was the largest budget line with the weakest accountability. Advertisers were pouring money into Connected TV with little more than proxy metrics to show for it. In 2026, however, advertisers expect performance from the system as a whole, not from any single channel in isolation. The key change is where responsibility lies. Advertisers no longer want to manage hundreds of levers, bids, segments, and pacing rules across different channels. They want to define objectives, set constraints and success metrics, then hold platforms accountable for delivering the results.
Historically, walled gardens taught the market that automation often comes with opacity – “trust us” black boxes that optimize behind closed doors. Viant is making the opposite case with its new Autonomous Outcomes solution: Viant’s AI enables autonomous execution across the open internet while preserving visibility into where media runs, how spend is allocated, and how results are achieved. In other words, automation doesn’t have to mean blind trust. Autonomous Outcomes isn’t just another optimization tool; it fundamentally shifts execution responsibility from human traders to an AI-driven system that plans, runs, and optimizes campaigns across the open internet. Advertisers simply declare their objectives and constraints, and the platform is accountable for delivering on those outcomes. This marks a dramatic change: instead of humans tweaking campaigns and hoping for the best, the system itself takes on the mandate of hitting the marketer’s goals – and it must transparently show how it’s doing so.
The Fragmenting Performance Market
The performance market is fragmenting along a predictable fault line – driven less by channel preferences and more by structural models. On one end of the spectrum are platforms where performance is tightly coupled to owned media and first-party data. Companies like Google, Meta, and Amazon deliver strong, repeatable performance inside their walled ecosystems because they control both the media and the data. Logged-in users, unlimited and abused supply, and closed-loop measurement make optimization very efficient. The tradeoff is portability: performance is most reliable when spend stays inside those environments. Marketers have limited ability to manage performance holistically across platforms because each walled garden optimizes in its own silo.
At the other end are platforms built to make a specific channel behave like performance media. In CTV, for example, companies such as tvScientific and MNTN focus on outcome-based buying and optimization within CTV. These approaches work well when TV is the primary lever and when success can be measured largely within that channel. The limitation is scope. Their performance logic doesn’t naturally extend across channels or coordinate spend across the full media mix. Optimization might be effective inside CTV, but it can’t easily balance TV with other channels under a unified strategy.
What advertisers are increasingly seeking sits between these two models. They want systems that operate across channels and optimize toward defined outcomes without the conflicts of interest that come from owning media or the narrow focus of single-channel tools. In this environment, independent and agnostic execution is essential – a platform that isn’t beholden to its own media can truly optimize across the open internet. However, autonomy alone isn’t enough. An autonomous cross-channel system can only succeed if it leverages superior data and signals; otherwise, it’s just optimizing on the same noisy, commoditized inputs as everyone else. This is where Viant’s approach comes into play, focusing on the quality of inputs feeding the AI.
Viant’s Approach: The Four Inputs for Autonomy
Most platforms talk about AI as if all the “smarts” live solely in the algorithm. In reality, outcomes are largely determined by what the system can see, what it’s allowed to act on, and what distortions are removed before a decision is made. Autonomy doesn’t fail because machines can’t optimize – it fails when machines are optimizing against noisy, commoditized, or compromised inputs. Viant’s approach starts earlier in the decision process, ensuring that the AI is working with the best possible information. Performance outcomes are shaped by four key inputs working together, which determine the quality of every downstream decision:
- Direct access to supply: Fewer intermediaries between the advertiser and the inventory preserve signal fidelity and allocate more of each dollar to working media. Cleaner auctions mean identity resolution is more reliable, contextual data is richer, and overall spend is more efficient. A direct path to supply ensures the system isn’t flying blind or losing valuable attributes along the way.
- Rigorous filtering of low-quality supply: Performance only exists if ads reach real humans. Much of the industry still treats fraud and low-quality placements as a reporting problem to be cleaned up after the fact. Viant treats it as a decisioning constraint from the start. By preemptively filtering out inventory that is fraudulent, non-viewable, or otherwise low-value, an autonomous system avoids wasting budget on impressions that could never drive outcomes. This “what not to buy” discipline protects the AI from chasing metrics on inventory that won’t move the needle for an advertiser.
- Robust identity resolution: Exposure without a connection to real people or households is wasted. Viant’s household-centric identity framework links identifiers, devices, individuals, and households into a unified view. This allows the system to understand reach and frequency across channels and to connect ad exposure to outcome events at a person/household level. Identity is the thread that ties the cross-channel journey together, ensuring the AI optimizes toward true incremental reach and results rather than siloed metrics.
- Differentiated data signals: Not all data is created equal. Commoditized datasets might maintain parity, but they won’t give you an edge. Unique, high-quality signals improve performance because they drive smarter decisions. Viant prioritizes inputs that materially change decisioning – for example, content-level intelligence or vertical-specific data that others don’t have or can’t easily use. These signals are valuable because they are scarce, defensible, and directly actionable. In a world where many platforms use the same inputs, having better data is often the decisive advantage. By feeding its decisioning engine with proprietary insights, the system can find opportunities and optimizations that generic DSPs will miss.
All four of these inputs feed continuous feedback loops within the platform. Measurement and outcome signals inform the AI so it can reallocate budget, adjust pacing, and tweak targeting on the fly. Viant’s decisioning system acts on this feedback to move spend efficiently, learn from each result, and improve performance over time. In essence, the quality of input data and signals gives the autonomous system a higher “ceiling” for performance – and the constant feedback ensures it keeps getting better the more it runs.
Accountable Performance in 2026
As we head into 2026, performance is being defined by outcomes above all else. Every tactic – budgets, bidding strategies, pacing adjustments, channel mixes – must justify itself by contributing to the advertiser’s bottom-line result. A few hallmarks of this new performance mindset:
- Outcome-based budgeting: Budgets are governed by cost-per-action or return-on-ad-spend guardrails. Instead of “We have $X to spend on CTV,” it’s “We will spend as long as we’re hitting $Y per acquisition or better.” Spend automatically scales up or down based on real performance against the goal.
- Incrementality is required: Advertisers demand proof that ads are driving incremental outcomes, not just cannibalizing organic sales or taking credit for what would have happened anyway. In 2026, merely showing attribution isn’t enough – platforms must demonstrate they’re delivering net new customers or sales that wouldn’t have occurred otherwise.
- Explainable results: Measurement must do more than confirm that results happened; it needs to explain why and how they happened. This means analytics that tie outcomes back to the tactics or segments that drove them. Marketers expect transparency into which creative, audience, or context delivered and what they can learn from it. Performance reports are evolving from simple dashboards into diagnostic tools.
- AI as the execution engine: AI is shifting from suggesting actions for humans to take, to directly executing decisions in real time. Instead of a planner adjusting bids based on a dashboard’s recommendation, the AI reallocates spend across campaigns and channels on its own, within the boundaries of the advertiser’s objectives. The role of the human shifts to setting the strategic goals and constraints, then overseeing the AI rather than micromanaging each lever.
- Simplicity over control: Perhaps counterintuitively, the platforms that win in this era won’t be those offering the most granular manual controls to advertisers – they’ll be the ones that know when to remove those controls. The ability to simplify the marketer’s job is now a competitive advantage. The best systems give transparency and levers when needed, but otherwise let the machine do the heavy lifting. In a world where every platform claims AI, the winners will prove it by letting marketers hand over the keys and delivering results.
The next phase of performance marketing will not be won by adding more controls, more elements, or more layers of abstraction. It will be won by systems willing to take accountability for outcomes across the open internet. Autonomy without transparency is just a black box. Intelligence without quality inputs is just noise. And performance that depends on complexity and hidden fees is not performance at all. Platforms that cannot simplify execution, explain results, and align incentives with advertisers will steadily lose relevance. The future belongs to systems that make accountability the product.