At A Glance
Agentic AI is moving faster through media planning than most organizations can define what it means for their business. Vendors are pitching it, conferences are built around it, and marketing teams feel pressure to move now. Before automation reallocates budget or reshapes a customer journey, marketers need three things in place: a shared definition of agentic AI, an identity foundation the system can trust, and independent measurement that keeps buying and performance separate.Agentic media planning has gone from emerging idea to assumed industry direction in a single cycle. The instinct to move fast on new technology makes sense. When a whole industry agrees on something quickly, though, the definitions, limits, and trade-offs that matter most tend to get skipped, and “agentic AI” still means very different things to different teams. Organizations building toward it without a shared definition, risk creating incompatible systems under the same label. There’s foundational work to do before automation takes the wheel.
What does agentic AI mean in media?
Much of the current enthusiasm around agentic AI conflates the technology with a specific tool. In many conversations, “agentic” has become shorthand for highly autonomous generative AI, as though large language models (LLMs) were the entire system rather than one component within it.
That framing understates what agentic systems are. Agentic systems are decision frameworks, not models. They coordinate predictive models, generative tools, and non-AI logic to move from signal to recommendation to action.
This pattern already exists outside media. Starbucks’ “Deep Brew” connects real-time purchase signals with personalized offers while informing what products stores stock and promote. The system coordinates customer demand with operational decisions across the business. Layering prompt engineering onto an LLM is a different thing entirely.
Without a shared definition of what “agentic” means for your organization, teams risk optimizing different visions of the same idea:
All may qualify as agentic, but without a shared definition, you can’t build toward the same goal. That’s why defining the problem statement comes first, not last.
Why do acting systems raise the stakes?
Here’s what sets agentic systems apart from earlier waves of automation: they don’t just inform decisions; they execute them. Programmatic has always operated within guardrails, executing and optimizing based on defined parameters. Agentic systems work more like navigation. They determine direction and weigh trade-offs against a wider set of context. When an agent reallocates budget mid-flight, suppresses an audience segment, or swaps creative, the quality of the identity and business logic behind that decision determines the outcome.
Here’s what sets agentic systems apart from earlier waves of automation: they don’t just inform decisions; they execute them.
Apply that framework to media. An agentic system operating on incomplete identity data or stale signals won’t slow down to account for the gap. It optimizes faster in the wrong direction. The coordination layer needs to weigh reach, incrementality, audience fatigue, and margin before it shifts spend, and that requires inputs that are accurate, current, and governed. Accountability and oversight aren’t optional anymore; they need to be built into the design from the start.
Watch our Curated Couch session on agentic media planning
Why should measurement stay separate from buying logic?
As autonomy grows in media systems, a structural problem emerges: the system shouldn’t measure its own performance.
When buying logic and measurement logic live in the same loop, performance numbers can look strong simply because they’re optimized to the system’s own signals. Success stops being defined externally and starts being defined internally. Agentic models trained on this pattern learn from the signals they favor, optimizing toward the outcomes they’re best equipped to observe. Performance looks good inside the system, and reflects a narrower view of what’s actually happening.
Independent measurement gets framed as friction on automation more often than not. In reality, it’s what lets automation to scale responsibly. Incorporating independent measurement signals into agentic decision-making, while maintaining meaningful separation from buying logic, produces an objective view of performance, one that can be compared consistently across systems rather than optimized within a single one. That distinction matters when you need to explain results to a client, stakeholder, or regulator.
AI as coordination layer, not an infrastructure replacement
Nothing here slows down automation. Independent measurement doesn’t add friction to execution but builds confidence in the outcomes those decisions generate. The organizations that will benefit most from agentic media won’t necessarily be the ones that adopt it fastest, but those that take the time to define where trusted autonomy creates measurable value and where human judgment needs to stay in the loop.
Before automation reallocates dollars or reshapes customer journeys, the data model has to hold up under scrutiny.
Trust in data, systems, and decision-making processes becomes the real constraint as agentic systems move from insight to action. That trust doesn’t come from autonomy alone, but it comes from governed data and decisions that can be explained, audited, and defended. Every brand that has operationalized automation at scale has learned the same lesson: before automation reallocates dollars or reshapes customer journeys, the data model has to hold up under scrutiny.
How does Experian help you build a defensible foundation?
Experian’s approach to AI in marketing starts from the same principle: intelligence is only as reliable as the foundation it’s built on.
We have spent decades building and governing data in regulated industries. That experience is what makes our foundation AI-ready: not just technically capable but built to withstand the scrutiny that responsible automation requires.
2026 State of advertising report
Our 2026 State of advertising report brings together perspectives from 14 leaders operating across key parts of the advertising ecosystem to show how shifts in AI, commerce media, healthcare, and more, are taking shape in practice.
Download our 2026 State of advertising report and hear from Scott Bender, Head of Publisher & Platform Partnerships, Newton Research, on how AI in advertising is reshaping planning, activation, and measurement.
The data foundation comes first
Agentic AI will reshape how media is planned and bought. That shift is happening now, and the opportunity is significant. But the advantage belongs to organizations that build on a foundation their systems, partners, and stakeholders can trust. Before automation runs, the inputs must be defensible, identity must be accurate, measurement must be independent, and the problem statement must be clear. Get those things right, and automation becomes more powerful. Skip them, and you’re simply moving faster in the wrong direction.
About the author
Budi Tanzi
SVP, Product, Experian
Budi Tanzi is the Senior Vice President of Product at Experian Marketing Services, overseeing all identity products. Prior to joining Experian, Budi worked at various stakeholders of the ad-tech ecosystem, such as Tapad, Sizmek, and StrikeAd. During his career, he held leadership roles in both Product Management and Solution Engineering. Budi has been living in New York for almost 11 years and enjoys being outdoors as well as sailing around NYC whenever possible.
FAQs
Agentic AI refers to systems that move from signal to recommendation to action with a high degree of autonomy. In media, this means an agent can reallocate budget, adjust audience targeting, or swap creative mid-flight based on real-time signals. Unlike standard programmatic systems that execute within predefined guardrails, agentic systems function more like a navigation layer, coordinating across data inputs, predictive models, and business rules to make directional decisions.
Data quality matters more as media automation becomes more autonomous because autonomous systems don’t slow down to account for bad inputs. If an agentic media system operates on incomplete identity data or stale audience signals, it optimizes faster toward the wrong outcomes. The quality of the identity foundation, the accuracy of audience attributes, and the reliability of measurement inputs all become more consequential, not less, as human oversight is reduced.
Programmatic executes and optimizes within defined parameters whereas agentic systems function at a higher level of abstraction, weighing reach, incrementality, fatigue, and margin before determining where and how to shift spend. Programmatic is the engine. Agentic AI, when built correctly, is the navigation layer that determines direction based on wider context and business objectives.
Measurement should be kept separate from buying logic in agentic systems because when a system measures its own performance, it optimizes toward outcomes it’s best equipped to observe. The definition of success becomes internal rather than external, and performance numbers may look strong while reflecting a narrower view of what’s working. Independent measurement, kept meaningfully separate from buying logic, produces results that can be compared across systems and explained to stakeholders without ambiguity.
Experian’s Offline Graph and Digital Graph provide a privacy-first identity foundation covering more than 250 million U.S. consumers and 4.2 billion digital IDs. Experian’s Marketing Data adds more than 5,000 attributes per audience, and Experian’s outcomes measurement keeps performance signals independent from buying logic.
Latest posts
To our valued customers and partners, it’s been an exciting week here at Tapad! As announced in a press release this morning, Tapad is now a member of the Experian family. We’re thrilled to continue to grow as a leader in identity resolution under the umbrella of a global expert in data, analytics and technology. Tapad and Experian are deeply connected by our commitment to serving the needs of our customers; and with a focus on quality of the data we provide, we have a common goal for the future of identity in the advertising ecosystem. As part of this announcement, we wanted to assure you, our valued customer, that we remain deeply committed to serving you today just as we always have. Nothing will change in your daily operations with Tapad. Experian immediately recognized that the success and growth of Tapad was directly tied to the strength and depth of its team members. As such, the acquisition will not result in any changes to day-to-day contacts at Tapad, or processes with weekly graph deliveries and other product support. Experian’s faith and investment in Tapad’s future and the future of identity resolution underscores what we’ve always believed our products could achieve and that we will be able to continue serving brands, advertisers, publishers, and the advertising and marketing ecosystem for years to come. On a personal note, I am excited to be transitioning my role as Chief Operating Officer of Tapad to the General Manager position of a global business that’s achieved exponential growth over the past several years; culminating in this strategic acquisition that will no doubt bring even more value to our customers in the future. We remain committed to open communication and welcome any questions you may have. Thank you,Mark Connon | General Manager, Tapad Contact us today
Addressable TV has been through a transformation in the past year. Streaming content has become the most coveted space for creators and advertisers with the rise of new apps and platforms; but the influx of stay-at-home orders around the country have shifted television viewership as we know it, and streaming apps are popping up in droves to take advantage. So, how can you? With no shortage of opportunities to advertise on addressable TV and CTV, how does it fit into the media mix? And furthermore, how can you attribute this household-level device into your overall strategy? Tying it all together Layering addressable TV within digital ad campaigns couldn’t be easier today — but applying the right targeting and cadence between all of your digital efforts; and tying them together in attribution takes the right kind of data. Marketers can use CTV identifiers coupled with other device identifiers available in The Tapad Graph to not only target impressions but also map addressable TVs within the consumer journey; and unify strategies between household decision makers to better personalize messaging. 1 The Trade Desk Q2 2020 Earnings Call Transcript, August 2020; 2 iSpot Report, via Deadline, July 2020; 3 Flixed.io, January 2020 Contact us today
For the past several years ad-tech defined the value of identity at the individual level; made possible by the evolution of data, technology and machine-learning. But, earlier this year COVID-19 set in motion many shifts in consumer digital behavior. The more we’ve been working and learning from home, using devices that are shared amongst an entire household, the more apparent it is that marketers need to shift their strategies to align with these changes. Did you know the average household owns eleven or more connected devices? And the longer we’ve been at home, the more these devices are shared by multiple individuals. If you’re looking for a few simple ways to evolve from an individual focused strategy to a household strategy, here’s a good place to start: Audience segmentation Traditionally, audiences are built with a narrow focus on a single user, and what known attributes about that individual or their brand engagement can be leveraged for a targeting strategy. Now that screens are being shared between multiple users in a home, how can you be sure you’re identifying them correctly, and thus, segmenting them in the right buckets for targeting? The key lies in the ability to connect those points through identity resolution. Using ad exposure from household level devices, followed by a second engagement from an individual within that household can indicate a user is a better candidate for purchase or conversion than others. So before you build audiences for targeting, you can qualify them at the household level for segmentation with more confidence. Example: An auto advertiser uses audience segments from a third party provider such as ‘auto intenders’ to target individuals with new pricing offers. They would continue retargeting these users, unaware that some are connected in the same household, and thus are probably not all in the market to actually get a new car. By bucketing users that share a common household device within this third party segment, they can hone in on which individuals are actually in-market for a car and evolve their strategy to be more effective. Targeting Retargeting, frequency capping and sequential messaging have always been meant for an individual user — the more they’re exposed to your brand in a personalized way, the more likely they are to take the desired action. But, have you considered that multiple users could have a shared initial exposure to your brand? Today, you can target a household of potential consumers on a shared device like a CTV, and employ those retargeting strategies based on that common initial exposure. Starting at the household level, means you can compare movement through the funnel between different individuals in that household, and tailor your targeting accordingly. Perhaps you realize only one person in that household will convert and you tailor messaging to them more frequently, while confidently suppressing the other individuals. Example: a CPG brand uses OTT advertising, but doesn’t incorporate it within their sequential strategy, because they consider it just a ‘brand awareness’ opportunity. By using OTT more strategically as a household level engagement, it can reveal which individuals within a household are more favorable towards a brand further down the funnel. So, you can spend impressions targeting those users, rather than wasting impressions on multiple individuals within the household. Measurement Measurement and attribution are imperative to understanding the path to purchase and making strategies more efficient over time. Often that efficiency involves adding or removing devices and channels from a targeting strategy based on their contribution to an action or conversion by an individual. This year we’re seeing addressable TV devices explode in use, which are shared at the household level. Even desktop computers are being used by more people in the home due to COVID-19. So, assuming a linear path of attribution by an individual is missing the full picture. Identity resolution can help you understand where messaging was more effective for some users in the household than others, and leverage that insight to continue more effective strategies in the future. Example: Without a household view, a direct-to-consumer brand would assume all interactions from one device would be coming from a single individual, and that could create a higher cost-per analysis. By incorporating the household level devices into attribution models, they can find efficiencies between touch points of multiple users, and learn how those split off into individual paths to conversion. Not only can this DTC create a more effective model, but they can use that model to create cost efficiencies in the future. Contact us today