The AI performance boom is real, but it’s trapped in search and social

Nearly 76% of advertisers see AI-driven performance gains, but most of the impact is still trapped inside Google and Meta. That’s according to research from Taboola titled "The agentic advantage in performance marketing: Securing incremental growth beyond search and social".

The study reveals that while agentic AI is already delivering strong results for performance marketers, its benefits remain heavily concentrated within walled gardens, leaving the open web structurally under-optimised despite clear advertiser demand.

At the same time, 82% of performance marketers say they are ready to adopt intelligent, goal-based campaign systems that can drive outcomes beyond search and social, signalling strong appetite for expansion beyond traditional channels. Despite this, budgets remain tightly anchored in search and social, with 74% of advertisers allocating at least a quarter of their performance spend to paid search and 67% doing the same for paid social. By contrast, the open web typically receives a moderate allocation.

Don't miss: AI shifts CMOs from marketing performance to growth accountability: Forrester 

The research also shows that at-scale adoption of AI campaign tools is still dominated by a small number of platforms. Google Performance Max is used at scale by 91% of respondents, while Meta Advantage+ is used by 88%. Outside of these ecosystems, adoption falls sharply. While TikTok’s Smart+ is widely being tested, open web campaign solutions are only used at scale by 36% of advertisers, despite 44% currently piloting them.

While intent is high, advertisers surveyed said expansion beyond search and social is being held back by operational constraints rather than lack of belief in performance. The single biggest barrier is difficulty integrating agentic AI tools into existing workflows (54%), followed by vendor complexity (74%) and fragmented measurement systems (71%) when it comes to scaling open web investment.

Brand safety concerns (54%) and limited internal resources (42%) further add to the challenge, highlighting that the issue is less about performance potential and more about operational scalability.

Despite these barriers, the demand signal is clear: 75% of marketers say finding incremental performance beyond search and social is very or extremely important, with urgency highest among senior decision-makers and high-spending advertisers.

Furthermore, the vast majority (80%) say they would immediately increase their ad spend on the open web if comparable agentic AI-powered solutions were available, with 86% willing to allocate up to a quarter of their performance marketing budget to the shift.

If equivalent AI-powered automation were available on the open web, 81% of advertisers say they would increase investment, with many indicating they would allocate up to a quarter of their performance budgets to the channel.

On average, respondents expect to allocate 24% of their performance budget to the open web under an AI-driven model, with at least 39% saying they would increase investment to 26% or more.

This suggests that while the open web is currently underutilised, it could become a materially larger performance channel if agentic AI systems reduce complexity and improve measurement parity with walled gardens.

The findings point to a structural imbalance in performance marketing: AI is already delivering measurable gains, but largely within closed ecosystems where automation is most mature.

As advertisers push for incremental growth beyond search and social, the next phase of performance marketing will likely hinge on whether open web environments can match the level of AI-driven optimisation, attribution clarity and workflow integration currently offered by Google and Meta. For now, advertisers are ready to expand, but the infrastructure is not. 

A broader pattern is also emerging beyond advertising. A McKinsey report found that while AI adoption is now widespread across Southeast Asia, most organisations are still struggling to translate usage into measurable business value.

The study notes that although most companies have begun using AI across at least one business function, a much smaller share have successfully scaled it into core operations. Even where adoption is higher in markets such as Singapore and Indonesia, many organisations remain in pilot or early-stage deployment, highlighting a gap between experimentation and full-scale integration.

Related articles: 
AI adoption still stuck at the starting line for most firms in Singapore
Brands struggle with AI disclosure as usage surges across marketing 
More companies miss revenue targets as AI and volatility reshape B2B growth