Agentic AI is quickly becoming one of the most discussed – and misunderstood – shifts in digital advertising. Much of the debate has focused on whether autonomous AI agents will actually replace search, social, or other existing performance channels.
But framing agentic commerce that way misses the point.
The reality is much more pragmatic. Agentic AI won’t replace performance media. Instead, it will add new layers of intent and discovery – and that means that our performance media strategies need to evolve with them.
At Criteo, we see agentic AI as the next phase of performance advertising: more complex, certainly – but also rich with new opportunities for measurable growth.
Agentic commerce is incremental, not disruptive
Consumers are already experimenting with agentic tools for both research and discovery, but this shouldn’t be seen as a wholesale behavioural reset. On the contrary it’s incremental.
Criteo research shows that while 40 per cent of US shoppers now use agentic assistants for product research, 96 per cent still engage across traditional touchpoints – search engines, social platforms, marketplaces, and brand or retailer sites – within the same journey.
And this tells us something important: agentic AI expands the places commerce happens, but it doesn’t replace the touchpoints that already drive results. For advertisers, the opportunity is capturing and activating these new intent signals – without abandoning the channels that still deliver.
From reactive signals to distributed intent
Performance marketing has historically been built on reactive inputs – keywords, site visits, retargeting pools, and last-click attribution. Agentic AI adds a new dimension: consumers expressing intent conversationally across multiple interfaces, sometimes before a click ever happens. Intent is now fragmented, contextual, and continuously evolving.
For performance teams, that means moving beyond channel-centric optimisation. The real challenge is now in recognising high-value intent earlier in the journey and activating media while it can still influence outcomes.
Why agentic AI raises the bar for performance execution
As discovery becomes more conversational, the tolerance for weak execution disappears. When product data is incomplete, pricing is outdated, or availability is unclear, performance simply falls apart – no matter how sophisticated the AI layer may be.
The move towards agentic commerce also exposes a potentially uncomfortable truth – that many performance strategies today rely on:
- Excessive frequency rather than relevancy
- Isolated optimisation rather than coordinated sequencing
- Volume over incrementality
In an agentic world, these shortcuts just won’t hold water. More impressions don’t equal better performance and, in fact, they often erode it.
Criteo’s role: Enabling agentic AI performance, not replacing it
Our focus at Criteo isn’t on building consumer-facing AI agents. Instead, we’re building tools to power the data and decisioning layer that makes agentic performance possible at scale.
By combining transactional data, a structured product catalogue, and real-time behavioural signals, we help advertisers move from “more targeting” to smarter activation – deciding when to engage, when to suppress, and how to sequence exposure based on conversion likelihood.
This isn’t black-box automation. Human oversight, transparency, and performance accountability remain essential. Agentic AI handles the complexity, but marketers stay in control.
Why ‘more AI’ doesn’t equal better performance
One of the most persistent misconceptions in digital advertising is that adding more AI automatically improves outcomes. In reality, performance improves when intelligence is applied with discipline.
Our experience shows that the best results come from:
- Reducing wasted impressions
- Prioritising high-intent audiences
- Sequencing creative based on user readiness
- Optimising for incremental growth, not surface-level metrics
Agentic AI reinforces this thinking. It moves performance from a scale-first model to a precision-first one.
What this means for advertisers in MEA
For advertisers in the MEA region – where performance budgets are under constant pressure – this transition has a few clear implications:
- Performance planning needs to move upstream. Waiting for the final click is no longer enough when intent is expressed earlier through conversational and AI-driven interfaces.
- Measurement must evolve with behaviour. Attribution models must reflect agent-mediated journeys – without losing focus on outcomes.
- Data quality and transparency become strategic performance levers. In an agentic environment, poor data reduces efficiency and undermines trust and results.
A more demanding era for performance advertising
Agentic AI doesn’t simplify performance marketing. It makes it more demanding.
It rewards advertisers and partners who combine strong data foundations, intelligent automation, and commercial accountability – and it exposes those still relying on blunt tactics or outdated optimisation models.
At Criteo, our role is clear: to help performance teams turn this growing complexity into clarity – and to ensure that, in an autonomous world, performance still performs.
By Gosia Wajchert, Managing Director MEA, Criteo.
