For years, the advertising industry has treated artificial intelligence primarily as a recommendation engine. AI systems could identify audiences, suggest budget allocations, or surface optimization opportunities, but the actual execution still relied heavily on humans manually adjusting campaigns across fragmented platforms.
That model may no longer be sufficient.
As digital advertising expands across social media, connected TV, display, and digital out-of-home channels, speed has become just as important as insight. According to Perion CEO Tal Jacobson, the industry is now entering a phase where AI systems are expected not only to recommend actions, but to execute them autonomously in real time.
From Recommendations To Execution
“The difference is that recommendation engines tell you what to do, but execution engines actually do it, at scale, in real time,” Jacobson explains. “Execution engines operate on optimization rules and real-time signals, removing humans from decision loops where speed matters. This framework enables autonomy to execute directly whenever there’s a clear optimization signal and measurable benefit. That’s the shift from planning to real-time action.”
Jacobson says the industry’s core bottleneck has fundamentally changed over time. “For years, the constraint in media buying was insight, what to do, who to target, how to measure,” he says. “As those layers matured, the constraint shifted to execution, how quickly and consistently the decision could actually be carried out.”
That shift is becoming more visible as marketers attempt to coordinate campaigns across increasingly fragmented ecosystems. Human teams still define strategy and creative direction, but many optimization decisions now happen too quickly for manual intervention to remain practical.
“It breaks at scale and speed,” Jacobson says of the assumption that humans must remain involved in every media decision. “Humans add real value when decisions are strategic, ambiguous, or require judgment about brand and context. Humans add less value when the decision is ‘shift two percent of budget because the conversion signal moved’, those decisions happen too often and too quickly for manual review to be useful. The practical answer is humans in the loop on strategy, AI in the loop on execution.”
The Cost Of Delayed Action
Jacobson argues that recommendation-only systems create inefficiencies that many organizations underestimate.
“Three risks,” he says. “First, latency cost: every minute between recommendation and action is performance you’ve already lost. Second, fragmentation cost: when each channel has its own recommendation system and a human stitches them together, you get inconsistent execution and conflicting signals.”
But he believes the larger issue is often hidden beneath the surface.
“Third, and the one that gets underestimated, opportunity cost: the recommendations a system surfaces are shaped by what it thinks a human can act on. The most valuable optimizations are often the ones that are too granular, too frequent, or too cross-channel for a human to ever execute. Those never even surface in a recommendation-only model.”
Perion’s answer to that challenge is Outmax, its AI-powered execution platform designed to directly manage budget allocation and pacing across platforms, including YouTube and Meta.
“Outmax provides multiple competitive advantages,” Jacobson says. “It executes in real time, giving brands speed and autonomy that traditional tools can’t match. It optimizes for brand KPIs, not platform KPIs, so your goals drive the execution, not vendor metrics.”
Why Cross-Channel Intelligence Matters
Jacobson believes the industry’s next major advantage will come not from isolated automation, but from systems capable of learning across campaigns and channels simultaneously.
“Because the industry built channel by channel,” he says when explaining why cross-campaign intelligence has remained elusive. “Every major platform optimized inside its own walls and treated cross-channel as someone else’s problem. The result is a stack where each channel is locally efficient and the system as a whole is globally inefficient.”
Perion’s broader infrastructure layer, Perion One, is designed to connect signals across CTV, DOOH, social, and display campaigns so that learnings compound over time instead of resetting with every campaign.
“The key word is compounds,” Jacobson says. “Every campaign Outmax executes is designed to contribute signals to a shared learning layer over time. The system isn’t starting from zero every time.”
That unified execution layer, he argues, reveals insights that single-channel optimization cannot detect.
“Single-channel optimization can’t see that exposure on CTV is reducing the cost of conversion on social, or that DOOH frequency is suppressing search intent,” Jacobson explains. “You only see those patterns when the same system is executing across all of them.”
The New Role Of Human Marketers
Despite the growing emphasis on automation, Jacobson does not see AI replacing marketers. Instead, he believes automation changes where human value is applied.
“It elevates the role,” he says. “Traders stop spending their time on the mechanical work, the bid adjustments, the pacing fixes, the cross-channel reconciliation, and start spending it on the strategic work, the work that actually requires human judgment.”
As execution increasingly becomes automated infrastructure, Jacobson believes the defining advantage for brands will shift toward strategic clarity rather than operational efficiency.
“When execution stops being a differentiator, the marketers who win are the ones with the sharpest view of who their customer is, what they want to say to them, and what outcome they’re optimizing for,” he says. “Automation rewards conviction. It punishes vagueness.”
:::tip
This story was distributed as a release by Jon Stojan under HackerNoon’s Business Blogging Program.
:::